VOLUME 120
FEBRUARY 2007
NUMBER 4
© 2007 by The Harvard Law Review Association
ARTICLES
SOSA, CUSTOMARY INTERNATIONAL LAW,
AND THE CONTINUING RELEVANCE OF ERIE
Curtis A. Bradley, Jack L. Goldsmith & David H. Moore
TABLE OF CONTENTS
I. INTRODUCTION ........................................................................................................................870
II. ERIE AND MODERN FEDERAL COMMON LAW ..............................................................874
A. Pre-Erie General Common Law.........................................................................................874
B. Erie v. Tompkins.................................................................................................................876
C. Post-Erie Federal Common Law .......................................................................................878
III. PRE-SOSA DEBATES REGARDING THE DOMESTIC STATUS OF CIL ........................881
A. CIL’s Pre-Erie Status ..........................................................................................................882
B. CIL’s Domestic Status in the Absence of Political Branch Authorization ..................885
C. Did the ATS Authorize Courts To Apply CIL as Federal Law?....................................887
D. Scope and Sources of CIL in ATS Litigation .................................................................888
IV. SOSA, THE ATS, AND THE MODERN POSITION .............................................................891
A. CIL’s Pre-Erie Status ..........................................................................................................892
B. Sosa and the ATS ................................................................................................................893
C. Scope and Sources of CIL in ATS Litigation..................................................................896
D. CIL’s Post-Erie Domestic Status in the Absence of Political Branch
Authorization........................................................................................................................902
1. Inconsistencies Between Sosa and the Modern Position ..........................................902
2. Counterarguments ..........................................................................................................907
V. A POST-SOSA APPROACH TO CIL AS FEDERAL COMMON LAW...................................910
A. Possible Jurisdictional Bases for CIL as Federal Common Law..................................911
1. Federal .............................................................................................................................911
2. Diversity ..........................................................................................................................914
3. Interstate and Admiralty Disputes...............................................................................915
B. Possible Substantive Bases for CIL as Federal Common Law .....................................919
1. Statutes ............................................................................................................................919
2. Treaties.............................................................................................................................921
3. Executive Branch Authorization..................................................................................922
C. CIL as Federal Common Law: Future Debates...............................................................924
1. Corporate Aiding and Abetting Liability....................................................................924
2. The War on Terrorism ...................................................................................................929
3. International and Foreign Sources in Constitutional Interpretation.......................933
VI. CONCLUSION ..........................................................................................................................935
869
SOSA, CUSTOMARY INTERNATIONAL LAW,
AND THE CONTINUING RELEVANCE OF ERIE
Curtis A. Bradley,∗ Jack L. Goldsmith∗∗ & David H. Moore∗∗∗
This Article analyzes the Supreme Court’s 2004 decision in Sosa v. Alvarez-Machain
against the backdrop of the post-Erie federal common law. The Article shows that,
contrary to the assertion of some commentators, Sosa did not embrace the “modern
position” that customary international law (CIL) has the status of self-executing federal
common law to be applied by courts without any need for political branch authorization
and, indeed, is best read as rejecting that position. Commentators who construe Sosa as
embracing the modern position have confounded the automatic incorporation of CIL as
domestic federal law in the absence of political branch authorization (that is, the
modern position) with the entirely different issue of whether and to what extent a
particular statute, the Alien Tort Statute (ATS), authorizes courts to apply CIL as
domestic federal law. The Article also explains how CIL continues to be relevant to
domestic federal common law despite Sosa’s rejection of the modern position. The
fundamental flaw of the modern position is that it ignores the justifications for, and
limitations on, post-Erie federal common law. As the Article shows, however, there are a
number of contexts in addition to the ATS in which it is appropriate for courts to
develop federal common law by reference to CIL, including certain jurisdictional
contexts not amenable to state regulation (namely, admiralty and interstate disputes), as
well as gap-filling and interpretation of foreign affairs statutes and treaties. The Article
concludes by considering several areas of likely debate during the next decade
concerning the domestic status of CIL: corporate aiding and abetting liability under the
ATS, application of CIL to the war on terrorism, and the use of foreign and
international materials in constitutional interpretation.
I. INTRODUCTION
T
he most contested issue in U.S. foreign relations law during the
last decade has been the domestic status of customary international law (CIL).1 In the mid-1990s, the conventional wisdom among
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
∗
∗∗
∗∗∗
Richard and Marcy Horvitz Professor of Law, Duke Law School.
Henry L. Shattuck Professor of Law, Harvard Law School.
Assistant Professor, University of Kentucky College of Law.
For their comments and suggestions, we thank Steve Burbank, Bobby Chesney, Richard
Fallon, Craig Green, Marty Lederman, Caleb Nelson, Gerald Neuman, Michael Ramsey, Paul
Stephan, Mark Weisburd, and participants in workshops held at the University of Chicago, Duke,
George Washington, Georgetown, and Northwestern law schools. For their excellent research
assistance, we thank Geoffrey Antell, Wells Bennett, Michael Fawcett, Josh Kreamer, Amy Osborne, Matt Perault, Joel Schellhammer, and Jason Spitalnick.
Professor Goldsmith served as an expert declarant with respect to some of the issues in this
Article in Bowoto v. Chevron Corp., No. C 99-02506 SI (N.D. Cal. Aug. 30, 2006).
1 CIL, historically referred to as part of the “law of nations,” is the law of the international
community that “results from a general and consistent practice of states followed by them from a
sense of legal obligation.” RESTATEMENT (THIRD) OF THE FOREIGN RELATIONS LAW OF
THE UNITED STATES § 102(2) (1987); see also Statute of the International Court of Justice art.
870
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
871
international law academics was that CIL had the status of selfexecuting federal common law to be “applied by courts in the United
States without any need for it to be enacted or implemented by Congress.”2 This “modern position” was criticized by so-called “revisionists” who argued that CIL had the status of federal common law only
in the relatively rare situations in which the Constitution or the political branches authorized courts to treat it as such.3
A number of modern position proponents argue that the Supreme
Court’s 2004 decision in Sosa v. Alvarez-Machain4 resolved this debate
in their favor. Dean Harold Koh, for example, contends that “all of
the . . . circuits have [embraced the modern position] (and now the
U.S. Supreme Court has as well, in the Alvarez-Machain case).”5 Professor Ralph Steinhardt claims that CIL “was and [after Sosa] remains
an area in which no affirmative legislative act is required to ‘authorize’ its application in U.S. courts.”6 Professor Martin Flaherty similarly maintains that Sosa’s “import is to confirm that international custom was part of judicially enforceable federal law even in the absence
of a statute.”7
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
38(1)(b), June 26, 1945, 59 Stat. 1055, 1060 (stating that international custom is a source of law
that can be applied by the International Court of Justice “as evidence of a general practice accepted as law”).
2 Louis Henkin, International Law as Law in the United States, 82 MICH. L. REV. 1555, 1561
(1984); see also Kadic v. Karadžic, 70 F.3d 232, 246 (2d Cir. 1995); In re Estate of Ferdinand E.
Marcos Human Rights Litig., 978 F.2d 493, 502 (9th Cir. 1992); RESTATEMENT (THIRD) OF THE
FOREIGN RELATIONS LAW OF THE UNITED STATES § 111 cmt. d, § 115 cmt. e; Lea Brilmayer, Federalism, State Authority, and the Preemptive Power of International Law, 1994 SUP.
CT. REV. 295, 295, 303–04, 332 n.109; Harold Hongju Koh, Commentary, Is International Law
Really State Law?, 111 HARV. L. REV. 1824, 1846–47 (1998).
3 See Curtis A. Bradley & Jack L. Goldsmith, Customary International Law as Federal Common Law: A Critique of the Modern Position, 110 HARV. L. REV. 815 (1997); A.M. Weisburd,
State Courts, Federal Courts, and International Cases, 20 YALE J. INT’L L. 1 (1995); see also
Phillip R. Trimble, A Revisionist View of Customary International Law, 33 UCLA L. REV. 665
(1986).
4 124 S. Ct. 2739 (2004).
5 Harold Hongju Koh, The Ninth Annual John W. Hager Lecture, The 2004 Term: The Supreme Court Meets International Law, 12 TULSA J. COMP. & INT’L L. 1, 12 (2004).
6 Ralph G. Steinhardt, Laying One Bankrupt Critique to Rest: Sosa v. Alvarez-Machain and
the Future of International Human Rights Litigation in U.S. Courts, 57 VAND. L. REV. 2241,
2259 (2004).
7 Martin S. Flaherty, The Future and Past of U.S. Foreign Relations Law, 67 LAW & CONTEMP. PROBS. 169, 173 (2004); see also Leila Nadya Sadat, An American Vision for Global Justice: Taking the Rule of (International) Law Seriously, 4 WASH. U. GLOBAL STUD. L. REV. 329,
342 (2005) (“[A] six-member majority of the Supreme Court rejected [the revisionist] view in Sosa
v. Alvarez-Machain.”); Recent Case, 119 HARV. L. REV. 1622, 1627 (2006) (“In Sosa, the Court
clearly rejected this revisionist argument, adopting the language of the predominant view.”); The
Supreme Court, 2003 Term—Leading Cases, 118 HARV. L. REV. 226, 453 (2004) (“Much of the majority’s analysis is consistent with the view that . . . all customary international law has been included within federal common law.”); cf. William S. Dodge, Bridging Erie: Customary International Law in the U.S. Legal System After Sosa v. Alvarez-Machain, 12 TULSA J. COMP. & INT’L
872
HARVARD LAW REVIEW
[Vol. 120:869
Both the pre- and post-Sosa debates largely turn on the implications of the Supreme Court’s seminal decision in Erie Railroad Co. v.
Tompkins.8 Modern position proponents tend to discount Erie’s relevance to the domestic status of CIL.9 Revisionists, by contrast, insist
that Erie is of central importance in determining whether and to what
extent CIL has the status of federal common law.
The debate over Erie’s relevance to the domestic status of CIL has
significant practical implications. If modern position proponents are
correct about Sosa, and CIL automatically has the status of federal
law, CIL would provide a basis for federal question jurisdiction, and
courts would be authorized to use CIL to preempt inconsistent state
law and possibly even to override executive branch action and some
federal legislation.10 These consequences would dramatically expand
the international human rights litigation permitted under the Sosa de–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
L. 87, 96–97 (2004) (arguing that Sosa endorses a particularized rather than wholesale incorporation of CIL into federal common law); Gerald L. Neuman, The Abiding Significance of Law in
Foreign Relations, 2004 SUP. CT. REV. 111, 132 (arguing that Sosa allows courts to “recogniz[e]
and incorporat[e] international norms, to the extent that they can be harmonized with other federal law”). But cf. David H. Moore, An Emerging Uniformity for International Law, 75 GEO.
WASH. L. REV. 1 (2006) (arguing that Sosa not only endorses the revisionist position, but also evidences the emergence of a uniform doctrine governing the status of both treaties and CIL in federal courts).
8 304 U.S. 64 (1938).
9 See, e.g., Koh, supra note 2, at 1831 (“Curiously, [revisionists] read Erie as effecting a near
complete ouster of federal courts from their traditional role in construing customary international
law norms.”); Gerald L. Neuman, Sense and Nonsense About Customary International Law: A
Response to Professors Bradley and Goldsmith, 66 FORDHAM L. REV. 371, 380 (1997) (“[T]he
Erie decision did not require that federal courts stop citing cases decided before 1938 and reinvent
federal common law from scratch.”); Jordan J. Paust, Customary International Law and Human
Rights Treaties Are Law of the United States, 20 MICH. J. INT’L L. 301, 308 (1999) (criticizing
revisionists for “their nearly obsessive focus” on Erie and Swift v. Tyson, 41 U.S. (16 Pet.) 1
(1842)); Beth Stephens, The Law of Our Land: Customary International Law as Federal Law After
Erie, 66 FORDHAM L. REV. 393, 397 (1997) (“[W]hile Erie rejected the general common law, it
upheld the federal courts’ power to develop common law in areas properly governed by federal
law, including international law.”).
10 For the claim that CIL creates federal jurisdiction and preempts state law, see, for example,
RESTATEMENT (THIRD) OF THE FOREIGN RELATIONS LAW OF THE UNITED STATES
§ 111(1) (1987), and Brilmayer, supra note 2, at 303. For the view that CIL binds the executive
branch, see RESTATEMENT (THIRD) OF THE FOREIGN RELATIONS LAW OF THE UNITED
STATES § 115 reporters’ note 4; Michael J. Glennon, Raising The Paquete Habana: Is Violation
of Customary International Law by the Executive Unconstitutional?, 80 NW. U. L. REV. 321, 324–
25 (1985); and Jules Lobel, The Limits of Constitutional Power: Conflicts Between Foreign Policy
and International Law, 71 VA. L. REV. 1071, 1116–20 (1985). For the view that CIL might trump
inconsistent prior federal law, see RESTATEMENT (THIRD) OF THE FOREIGN RELATIONS LAW
OF THE UNITED STATES § 115 reporters’ note 4, and Louis Henkin, The Constitution and
United States Sovereignty: A Century of Chinese Exclusion and Its Progeny, 100 HARV. L. REV.
853, 876–77 (1987). We should emphasize that not every proponent of the modern position embraces all of these propositions, although there is significant academic support for each of them,
and for the more general claim that all of CIL is federal law. See generally Bradley & Goldsmith,
supra note 3, at 817 nn.3–4, 837 nn.150–51.
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
873
cision and would provide a vehicle by which U.S. citizens could challenge the actions of their government (state and federal) based on
evolving CIL. These consequences might have particularly significant
implications for challenges to executive action in the war on terrorism.
In this Article, we hope to make three contributions to the debate
about CIL’s domestic status after Sosa. First, we attempt to focus the
debate more directly on Erie and its implications for modern federal
common law. Like the Court in Sosa (but unlike many proponents of
the modern position), we believe that Erie is centrally relevant to the
current status of CIL in U.S. courts. Any theory of the domestication
of CIL as federal common law must be consistent with Erie’s basic
premises, and in this Article we attempt to flesh out the implications of
those premises for domesticating CIL.
Second, we examine the emerging claim that Sosa constitutes an
endorsement of the modern position that CIL is incorporated wholesale into the U.S. legal system as federal common law. As we show,
the decision in Sosa cannot reasonably be read as embracing the modern position and, indeed, is best read as rejecting it. Commentators
who construe Sosa as embracing the modern position have confounded
the automatic incorporation of CIL as domestic federal law in the absence of congressional authorization (that is, the modern position) with
the entirely different issue of whether and to what extent a particular
statute, the Alien Tort Statute11 (ATS), authorizes courts to apply CIL
as domestic federal law. The ATS, which was first enacted in 1789,
grants federal district courts jurisdiction over “any civil action by an
alien for a tort only, committed in violation of the law of nations or a
treaty of the United States.”12 The Court in Sosa held that the ATS
authorized federal courts to recognize federal common law causes of
action for a narrow class of CIL violations. The Court based this conclusion on what was, in effect, a translation of the specific intentions of
the ATS framers to the regime of post-Erie federal common law. As
we show, the Court’s analysis would be superfluous if it agreed with
the modern position. Moreover, the Court’s reasoning and conclusions
on a number of points simply cannot be reconciled with the modern
position.
Third, we explain how CIL continues to be relevant to domestic
federal common law despite Sosa’s rejection of the modern position.
The fundamental flaw of the modern position is that it ignores the justifications for, and limitations on, post-Erie federal common law. As
we show, however, there are a number of contexts in addition to the
ATS in which it is appropriate for courts to develop federal common
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
11
12
28 U.S.C. § 1350 (2000).
Id.
874
HARVARD LAW REVIEW
[Vol. 120:869
law by reference to CIL. These include certain jurisdictional contexts
— namely, admiralty and interstate disputes — and the gap-filling and
interpretation of foreign affairs statutes and treaties. In short, rejection of the modern position does not entail a rejection of the judicial
domestication of CIL, but at the same time courts can domesticate
CIL only in accordance with the requirements and limitations of postErie federal common law — limitations that, as we explain, were reaffirmed by the Court in Sosa.
This Article proceeds as follows. Part II describes the basic principles of Erie and its implications for modern federal common law. Part
III distinguishes and describes the various existing debates concerning
the domestic status of CIL and shows how Erie is central to each of
these debates. Part IV analyzes the Sosa decision, explains its implications for the existing debates, and shows how it is best read as rejecting the modern position. Part V considers some of the many ways in
which CIL can, consistent with Erie, inform the development of federal common law in the U.S. legal system even after rejection of the
modern position. It also discusses several likely areas of debate concerning the domestic status of CIL during the next decade.
II. ERIE AND MODERN FEDERAL COMMON LAW
This Part briefly describes the general common law framework that
existed prior to Erie, the Court’s justifications in Erie for rejecting
that framework, and the contours of the post-Erie federal common
law. We do not attempt here to offer new insights about these widely
discussed subjects; our goal is merely to remind readers of certain settled propositions that, as we explain in Parts III and IV, are relevant to
debates over the domestic status of CIL.
A. Pre-Erie General Common Law
Before Erie, federal and state courts in civil cases applied a body of
law that came to be known as “general common law.”13 They “resorted to [general common law] to provide the rules of decision in particular cases without insisting that the law be attached to any particular sovereign.”14 As Justice Holmes would eventually describe and
criticize it, general common law was “a transcendental body of law
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
13 See generally Bradford R. Clark, Federal Common Law: A Structural Reinterpretation, 144
U. PA. L. REV. 1245, 1279–85 (1996); William A. Fletcher, The General Common Law and Section
34 of the Judiciary Act of 1789: The Example of Marine Insurance, 97 HARV. L. REV. 1513 (1984);
Stewart Jay, Origins of Federal Common Law (pt. 2), 133 U. PA. L. REV. 1231 (1985). Early in
U.S. history, federal courts also applied a common law of crimes, but the Supreme Court disallowed this practice in the early 1800s. See United States v. Coolidge, 14 U.S. (1 Wheat.) 415
(1816); United States v. Hudson & Goodwin, 11 U.S. (7 Cranch) 32 (1812).
14 Fletcher, supra note 13, at 1517.
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
875
outside of any particular State but obligatory within it unless and until
changed by statute.”15
Courts did not view general common law as having the status of
federal law. They did not consider it part of the “Laws of the United
States” within the meaning of the Supremacy Clause,16 and claims
arising under it did not fall within either Article III or statutory federal question jurisdiction.17 Federal court interpretations of general
common law were not binding on state courts, and the two court systems sometimes adopted differing interpretations of this law.18
A famous early application of general common law occurred in
Swift v. Tyson,19 in which the Supreme Court applied “principles established in the general commercial law,” rather than New York state
court decisions, to resolve a commercial dispute concerning the validity
of an assignment of a negotiable instrument, even though the assignment had occurred in New York.20 In support of its decision, the
Court reasoned that the Rules of Decision Act,21 which requires federal courts to apply the “laws of the several states”22 in cases not governed by the Constitution, treaties, or federal statutes, applies only to
“the positive statutes of the state,” and not to state court decisions on
“questions of a more general nature.”23
Although relatively uncontroversial for much of the nineteenth century, the general common law regime became contested in the late
nineteenth and early twentieth centuries as courts began to apply it to
a broader array of cases. One of many criticisms of this regime was
that it allowed litigants to forum shop between federal and state courts
for the most favorable interpretation of the general common law. A
notorious example of such forum shopping occurred in Black & White
Taxicab & Transfer Co. v. Brown & Yellow Taxicab & Transfer Co.24
In that case, a Kentucky taxicab company reincorporated in Tennessee
so that it could sue another Kentucky taxicab company under diversity
jurisdiction for interference with an exclusive contract with a railroad
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
15 Black & White Taxicab & Transfer Co. v. Brown & Yellow Taxicab & Transfer Co., 276
U.S. 518, 533 (1928) (Holmes, J., dissenting). Justice Holmes argued elsewhere that, instead of this
“brooding omnipresence in the sky,” the common law should in fact be understood as “the articulate voice of some sovereign or quasisovereign that can be identified.” S. Pac. Co. v. Jensen, 244
U.S. 205, 222 (1917) (Holmes, J., dissenting).
16 U.S. CONST. art. VI, cl. 2.
17 See Fletcher, supra note 13, at 1521–25.
18 RESTATEMENT (THIRD) OF THE FOREIGN RELATIONS LAW OF THE UNITED STATES
ch. 2, introductory note, at 41 (1987).
19 41 U.S. (16 Pet.) 1 (1842).
20 Id. at 18.
21 28 U.S.C. § 1652 (2000).
22 Id.
23 Swift, 41 U.S. (16 Pet.) at 18.
24 276 U.S. 518 (1928).
876
HARVARD LAW REVIEW
[Vol. 120:869
and thereby obtain a federal court ruling concerning the validity of the
contract. The Supreme Court upheld the existence of diversity jurisdiction and affirmed judgment for the plaintiff, even though Kentucky
courts would have found the exclusive railroad contract to be invalid.
The Court in Black & White Taxicab explained its divergence from
the state decisions as follows:
For the discovery of common law principles applicable in any case, investigation is not limited to the decisions of the courts of the State in which
the controversy arises. State and federal courts go to the same sources for
evidence of the existing applicable rule. The effort of both is to ascertain
that rule.25
Justice Holmes, along with two other Justices, dissented. He argued
that the general common law regime rested on the “fallacy”26 that law
can exist without some definite authority behind it. Once it is recognized that state court application of general common law derives its
authority from the state and thus is in effect state law, he reasoned, the
practice of federal courts declining to follow that law in diversity cases
should be seen as “an unconstitutional assumption of powers by the
Courts of the United States.”27
B. Erie v. Tompkins
The specific issue in Erie was what law a federal court sitting in
diversity should apply to determine the tort duties that a railroad
owed to someone walking along the railroad’s tracks. In concluding
that state law should be applied, the Court held that “[e]xcept in matters governed by the Federal Constitution or by Acts of Congress, the
law to be applied in any case is the law of the State.”28 The Court further made clear that, henceforth, “[t]here is no federal general common
law.”29
The Court in Erie determined that, contrary to its holding in Swift
v. Tyson, the “laws of the several states” referenced in the Rules of Decision Act included judge-made law. Importantly, the Court explained
that if Swift had involved only a mistaken statutory interpretation, the
Court would have been hesitant to overrule it. But the Court concluded that the “unconstitutionality” of the general common law regime “ha[d] . . . been made clear and compel[led] [the Court]” to abandon it.30
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
25
26
27
28
29
30
Id. at 529–30.
Id. at 532 (Holmes, J., dissenting).
Id. at 533.
Erie R.R. Co. v. Tompkins, 304 U.S. 64, 78 (1938).
Id.
Id. at 77–78. In addition to relying on constitutional grounds, the Court noted the many
practical “defects, political and social,” that attended the general common law regime, including
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
877
In its constitutional analysis, the Court endorsed Justice Holmes’s
positivist argument that “law in the sense in which courts speak of it
today does not exist without some definite authority behind it.”31 As a
result, the Court denied what it called the “fallacy” that there is a
“transcendental body of law outside of any particular State but obligatory within it unless and until changed by statute.”32 This means, explained the Court, that the common law of a state “is not the common
law generally but the law of that State existing by the authority of that
State.”33
The Court connected this positivist approach to common law with
principles of federalism. The Court explained that “Congress has no
power to declare substantive rules of common law applicable in a
State whether they be local in their nature or ‘general,’ be they commercial law or a part of the law of torts,” and that “no clause in the
Constitution purports to confer such a power upon the federal
courts.”34 It concluded from these premises that disregard of state
court decisions on commercial and tort law questions “invaded rights
which . . . are reserved by the Constitution to the several States.”35
At the time of Erie, the Court’s constitutional analysis rested on
both congressional and judicial incapacity to make common law rules
on matters reserved to the states. The Court’s observation about limited congressional power became less important as the Supreme Court
approved significant expansions of legislative power in the post-Erie
period. As a result, the Court’s reasoning about constitutional limitations on the federal government’s power to invade state rights evolved
into an argument about limitations on the federal judiciary. As Professor Thomas Merrill notes, “the federalism principle identified by Erie
still exists but has been silently transformed from a general constraint
on the powers of the federal government into an attenuated constraint
that applies principally to one branch of that government — the federal judiciary.”36
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
the unfairness of allowing out of state plaintiffs to choose whether a case would be heard in state
or federal court based on which court had a more favorable view of the general common law. Id.
at 74–75.
31 Id. at 79 (quoting Black & White Taxicab, 276 U.S. at 533 (Holmes, J., dissenting)).
32 Id. (quoting Black & White Taxicab, 276 U.S. at 533 (Holmes, J., dissenting)) (internal quotation mark omitted).
33 Id.
34 Id. at 78.
35 Id. at 80.
36 Thomas W. Merrill, The Common Law Powers of Federal Courts, 52 U. CHI. L. REV. 1, 15
(1985); see also Wallis v. Pan Am. Petroleum Corp., 384 U.S. 63, 68 (1966) (“It is by no means
enough [to justify federal common law–making] that . . . Congress could under the Constitution
readily enact a complete code of law governing [the subject matter of the case]. Whether latent
federal power should be exercised to displace state law is primarily a decision for Congress.”);
George Rutherglen, Reconstructing Erie: A Comment on the Perils of Legal Positivism, 10
878
HARVARD LAW REVIEW
[Vol. 120:869
C. Post-Erie Federal Common Law
Although there are statements in Erie suggesting that the only
common law that federal courts can apply is the common law of the
states, Erie in fact gave birth to the development of a new common
law in the federal courts.37 This “federal common law” is genuine federal law that binds the states under the Supremacy Clause and potentially establishes Article III and statutory “arising under” jurisdiction.38
The Supreme Court has never provided a comprehensive explanation of its approach to federal common law after Erie, and it has sometimes simply referred to the fact that there are “enclaves” of federal
common law encompassing issues of national importance.39 Nonetheless, certain basic parameters of post-Erie federal common law can be
discerned from Erie itself and the various post-Erie federal common
law decisions: it derives its authority from extant federal law, it is interstitial, and it must be tailored to the policy choices reflected in its
federal law sources.
First, because “the federal lawmaking power is vested in the legislative, not the judicial, branch of government,”40 federal common law
must be grounded in extant federal law: the Constitution, a federal
statute, or a treaty. It is this grounding in a federal law source that allows federal common law to have the status of preemptive law under
the Supremacy Clause. Recall that Erie insisted that all law applied
by federal courts must derive from a domestic sovereign source and
thus must be either federal law or state law. Erie’s holding — that
state law governed in diversity cases — followed from these premises
because there was no basis in the Constitution or any federal statute
for federal courts to develop their own law in such cases. From these
same premises emerged the basic animating principle of post-Erie fed–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
CONST. COMMENT. 285, 288 (1993) (“Any major extension of federal power must find its source
in the Constitution or in a federal statute, not in the common law decisions of federal judges
alone.”).
37 See Henry J. Friendly, In Praise of Erie — and of the New Federal Common Law, 39
N.Y.U. L. REV. 383, 405–07 (1964).
38 See, e.g., Illinois v. City of Milwaukee, 406 U.S. 91, 100 (1972) (“[Section] 1331 jurisdiction
will support claims founded upon federal common law as well as those of a statutory origin.”); see
also Merrill, supra note 36, at 6–7 (distinguishing federal common law from general common law).
Federal common law is often still “general” in the sense that its content is derived from general
principles or practice. See Caleb Nelson, The Persistence of General Law, 106 COLUM. L. REV.
503 (2006). But it is not general in the sense of applying across entire fields of law, and it has federal law status, rather than general law status, in U.S. courts. Even the modern federal common
law of admiralty is applied in an interstitial way. See infra pp. 918–19.
39 See Banco Nacional de Cuba v. Sabbatino, 376 U.S. 398, 426 (1964) (noting that “there are
enclaves of federal judge-made law which bind the States”); see also Tex. Indus., Inc. v. Radcliff
Materials, Inc., 451 U.S. 630, 641 (1981) (recounting areas in which the Court has approved of
federal common law).
40 Nw. Airlines, Inc. v. Transp. Workers Union, 451 U.S. 77, 95 (1981).
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
879
eral common law: when the common law being developed by federal
courts did have some federal law basis, then it would have the status
of truly federal law.
The Supreme Court has not always explicitly articulated this requirement of a federal law source. Nevertheless, the reasoning in even
its most expansive federal common law decisions typically has reflected this requirement. In Clearfield Trust Co. v. United States,41 for
example, the Court held that the rights and duties of the United States
government concerning federal checks must be governed by federal
common law. The Court reasoned that, because the government’s issuance of the check in question stemmed from its constitutional powers and was for services performed under a federal statute, the federal
common law rights and duties associated with the check “find their
roots in the same federal sources.”42 To take another example, in
Banco Nacional de Cuba v. Sabbatino43 the Court held that the act of
state doctrine, pursuant to which courts assume the validity of foreign
government acts taken within their territory, is a rule of federal common law binding on the states.44 The Court grounded its decision in
two federal law sources: “‘constitutional’ underpinnings” relating to
the separation of powers in conducting U.S. foreign relations,45 which
the Court held “must be treated exclusively as an aspect of federal
law,”46 and numerous federal constitutional and statutory provisions
suggesting that sensitive foreign relations questions are exclusive federal concerns.47
While there is much scholarly debate about the proper contours of
federal common law, there is widespread agreement that federal common law must be grounded in a federal law source. For example, Professor Merrill advocates a restrictive approach to federal common law,
whereby there would have to be a showing that the federal common
law rule “can be derived from the specific intentions of the draftsmen
of an authoritative federal text.”48 By contrast, Professor Martha Field
argues for a broader approach, maintaining that the development of
federal common law is appropriate so long as the court can “point to
a federal enactment, constitutional or statutory, that it interprets as authorizing the federal common law rule.”49 The key point for present
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
41
42
43
44
45
46
47
48
49
318 U.S. 363 (1943).
Id. at 366.
376 U.S. 398.
See id. at 424–27.
Id. at 423–24.
Id. at 425.
See id. at 425–26 & n.25.
Merrill, supra note 36, at 47.
Martha A. Field, Sources of Law: The Scope of Federal Common Law, 99 HARV. L. REV.
881, 887 (1986); see also Larry Kramer, The Lawmaking Power of the Federal Courts, 12 PACE L.
880
HARVARD LAW REVIEW
[Vol. 120:869
purposes is that despite disagreement over how it is to be applied,
there is general agreement on the requirement of a federal law source.
Even Judge Henry Friendly, a particularly enthusiastic supporter
of federal common law, tied federal common law to congressional
intent.50
Second, the post-Erie federal common law must be interstitial; that
is, courts are to develop it only in retail fashion to fill in the gaps, or
interstices, of federal statutory or constitutional regimes. This requirement follows from the Court’s reasoning in Erie that “[f]ederal
courts, unlike state courts, are not general common-law courts and do
not possess a general power to develop and apply their own rules of
decision.”51 As Justice Holmes famously noted in a pre-Erie dissent
anticipating post-Erie federal common law, “judges do and must legislate, but they can do so only interstitially; they are confined from molar to molecular motions.”52
Third, when developing federal common law, courts must act consistently with the policy choices reflected in extant federal law. As Justice Jackson explained, “[f]ederal common law implements the federal
Constitution and statutes, and is conditioned by them.”53 This requirement follows from the fact that federal common law is a derivative form of lawmaking rather than an independent judicial power to
make policy decisions.54 As a result, “a federal court’s primary responsibility in deriving an appropriate federal common law rule is to attempt to give effect to the underlying federal policy found in federal
statutes, the Constitution, or another federal text.”55
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
REV. 263, 288 (1992) (“[F]ederal judges must wait for Congress to take the first step. Once Congress has acted, however, federal courts can make any common law ‘necessary and proper’ to implement the statute.”).
50 See Friendly, supra note 37, at 407 (“Just as federal courts now conform to state decisions on
issues properly for the states, state courts must conform to federal decisions in areas where Congress, acting within powers granted to it, has manifested, be it ever so lightly, an intention to that
end.”); see also id. at 422 (noting that “state courts must follow federal decisions on subjects
within national legislative power where Congress has so directed”).
51 City of Milwaukee v. Illinois, 451 U.S. 304, 312 (1981) (citing Erie R.R. Co. v. Tompkins,
304 U.S. 64, 78 (1938)); see also United States v. Standard Oil Co., 332 U.S. 301 (1947).
52 S. Pac. Co. v. Jensen, 244 U.S. 205, 221 (1917) (Holmes, J., dissenting). For post-Erie decisions in which the Court emphasized that federal common law is an interstitial judicial lawmaking power, see Kamen v. Kemper Financial Services, Inc., 500 U.S. 90, 98 (1991); United States v.
Kimbell Foods, Inc., 440 U.S. 715, 727 (1979); and United States v. Little Lake Misere Land Co.,
412 U.S. 580, 593 (1973).
53 D’Oench, Duhme & Co. v. FDIC, 315 U.S. 447, 472 (1942) (Jackson, J., concurring).
54 Professor Stephen Burbank argues that the proposition also follows from the Rules of Decision Act. See Stephen B. Burbank, Interjurisdictional Preclusion, Full Faith and Credit and
Federal Common Law: A General Approach, 71 CORNELL L. REV. 733 (1986); cf. Merrill, supra
note 36, at 31–32 (“The phrase ‘require or provide’ [in the Rules of Decision Act] is broad enough
to embrace at least some interpretation of federal texts, and thus to support the creation of some
federal common law.”).
55 Paul Lund, The Decline of Federal Common Law, 76 B.U. L. REV. 895, 968 (1996).
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
881
The Supreme Court has noted this requirement even in its broadest
assertions of federal common law. For example, in explaining the propriety of developing federal common law to resolve collective bargaining disputes brought under the Labor Management Relations Act of
1947,56 the Court stated that this law was to be “fashion[ed] from the
policy of our national labor laws” and that some federal common law
rules would “lie in the penumbra of express statutory mandates,”
whereas others “[would] lack express statutory sanction but [would] be
solved by looking at the policy of the legislation and fashioning a remedy that will effectuate that policy.”57 As Professors Peter Westen and
Jeffery Lehman explain, “[f]ederal common law is measured by the
same standard of validity as federal statutory interpretation; the measure in each case is whether the law as declared by the courts is consistent with prevailing legislative policy.”58
Consistent with these three principles, the Supreme Court has
stated that the instances in which it is appropriate to develop federal
common law are “few and restricted.”59 In the last fifteen years or so
in particular, the Court has been emphatic about the exceptional nature of federal common law.60 The Court has also adopted a restrictive approach in recent years to the judicial recognition of private
rights of action under federal statutes and the Constitution, which can
be seen as a remedial form of federal common law.61
III. PRE-SOSA DEBATES REGARDING
THE DOMESTIC STATUS OF CIL
The Sosa case concerned actions taken by the U.S. Drug Enforcement Agency (DEA). In 1990, the DEA recruited Sosa and other
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
56
57
ch. 120, 61 Stat. 136 (codified in scattered sections of 29 U.S.C.).
Textile Workers Union v. Lincoln Mills, 353 U.S. 448, 456–57 (1957); see also Kimbell Foods,
440 U.S. at 738 (“[I]n fashioning federal principles to govern areas left open by Congress, our
function is to effectuate congressional policy.”); Wallis v. Pan Am. Petroleum Corp., 384 U.S. 63,
69 (1966) (“If there is a federal statute dealing with the general subject, it is a prime repository of
federal policy and a starting point for federal common law.”).
58 Peter Westen & Jeffery S. Lehman, Is There Life for Erie After the Death of Diversity?, 78
MICH. L. REV. 311, 336 (1980); see also Kramer, supra note 49, at 287–88.
59 Wheeldin v. Wheeler, 373 U.S. 647, 651 (1963).
60 See, e.g., Atherton v. FDIC, 519 U.S. 213, 218 (1997) (“‘Whether latent federal power should
be exercised to displace state law is primarily a decision for Congress,’ not the federal courts.”
(quoting Wallis, 384 U.S. at 68)); O’Melveny & Myers v. FDIC, 512 U.S. 79, 87 (1994) (observing
that the instances in which it is appropriate to create a federal common law rule are “few and restricted” (quoting Wheeldin, 373 U.S. at 651)). As we explain in this Article, Sosa continues this
trend.
61 See, e.g., Corr. Servs. Corp. v. Malesko, 534 U.S. 61, 66–74 (2001) (explaining why Bivens
remedies for constitutional torts should not be extended to claims against private entities); Alexander v. Sandoval, 532 U.S. 275, 286 (2001) (“Like substantive federal law itself, private rights of
action to enforce federal law must be created by Congress.”).
882
HARVARD LAW REVIEW
[Vol. 120:869
Mexican nationals to abduct Alvarez-Machain, a Mexican national,
and transport him from Mexico to the United States to stand trial for
his alleged involvement in the torture and murder of a DEA agent in
Mexico.62 After he was acquitted, Alvarez-Machain sued Sosa under
the ATS, alleging that Sosa had violated a CIL prohibition on arbitrary arrest.63 Sosa, and the U.S. government acting as amicus curiae,
argued that the ATS was simply a jurisdictional statute that did not
create a cause of action for violations of CIL. Alvarez-Machain, by
contrast, argued that the ATS did create CIL causes of action, including a cause of action for arbitrary arrest.64
The legal controversy in Sosa arose against the background of debates, in the courts and the academy, about four distinct issues concerning CIL’s status in the domestic legal system. Since many commentators interpreting Sosa have confused the decision’s resolution of
one of these issues with its resolution of others, it is important to distinguish the issues carefully and to understand their relationship to one
another. The first debate concerns the historical status of CIL in the
U.S. legal system prior to Erie v. Tompkins during the era of “general
common law.” The second debate concerns CIL’s domestic legal
status, following Erie, in the absence of some authorization by the political branches to apply CIL as federal law. The third debate concerns whether the ATS or some related statutory enactment authorizes
federal courts to apply CIL as federal law consistent with the usual
requirements of post-Erie common law creation. The final debate
concerns the scope of the CIL that can be applied in ATS litigation
and, relatedly, how this CIL is to be identified. As we explain, Erie is
central to each of these debates.
A. CIL’s Pre-Erie Status
It is uncontroversial that, during the period prior to Erie, federal
courts often applied CIL (which they referred to as part of the “law of
nations”) without requiring authorization from the federal political
branches. Before Sosa, courts and scholars disagreed about whether
CIL so applied had the status of genuinely federal law or whether it
had the status of nonfederal general common law. Some proponents of
the modern position argued that CIL was federal law and thus escaped
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
62
63
Sosa v. Alvarez-Machain, 124 S. Ct. 2739, 2746–47 (2004).
Alvarez-Machain also sued the U.S. government under the Federal Tort Claims Act
(FTCA), 28 U.S.C. §§ 1346(b), 2401(b), 2671–2680 (2000). See Sosa, 124 S. Ct. at 2747. Although
the Ninth Circuit held that the government was liable for false arrest, the Supreme Court concluded that the government was immune from suit and ordered dismissal of Alvarez-Machain’s
FTCA claim. See id.
64 Sosa, 124 S. Ct. at 2754–55.
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
883
Erie’s abolition of general common law. Revisionists claimed that CIL
was general common law that fell squarely within Erie’s scope.
The debate turned in part on the meaning of certain historic statements concerning the domestic status of CIL. In the famous Paquete
Habana decision,65 the Supreme Court stated that CIL “is part of our
law, and must be ascertained and administered by the courts of justice
of appropriate jurisdiction, as often as questions of right depending
upon it are duly presented for their determination.”66 Similarly, the
Supreme Court stated in an earlier decision that it was “bound by the
law of nations which is a part of the law of the land.”67 And some of
the constitutional Founders, speaking later as judges, maintained that
the law of nations was part of U.S. law.68
Citing these and similar statements, some courts and scholars argued that CIL historically had the status of federal law in the U.S. legal system.69 Dean Koh claimed, for example, that CIL’s status as
federal law has been established since “the beginning of the Republic”
and reflects “a long-accepted, traditional reading of the federal courts’
function.”70 Supporters of this view interpreted phrases like “law of
the land,” “law of the United States,” and “our law” as references to
federal law that preempts state law and creates a basis for federal
question jurisdiction. In addition, they claimed that viewing CIL as
nonfederal law, and thus as not judicially enforceable against the
states, would have been inconsistent with the Founders’ welldocumented desire to ensure that states complied with international
law. The Second Circuit in Filartiga v. Pena-Irala71 (the first decision
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
65
66
67
The Paquete Habana, 175 U.S. 677 (1900).
Id. at 700.
The Nereide, 13 U.S. (9 Cranch) 388, 423 (1815); see also 11 Op. Att’y Gen. 297, 299–300
(1865); 1 Op. Att’y Gen. 566, 570 (1822); 1 Op. Att’y Gen. 26, 27 (1792); ALEXANDER HAMILTON & JAMES MADISON, LETTERS OF PACIFICUS AND HELVIDIUS ON THE PROCLAMATION OF NEUTRALITY OF 1793, at 15 (Richard Loss ed., 1976).
68 See Chisholm v. Georgia, 2 U.S. (2 Dall.) 419, 474 (1793) (John Jay); United States v. Worrall, 28 F. Cas. 774, 778 (C.C.D. Pa. 1798) (No. 16,760) (Samuel Chase and Richard Peters, Jr.);
United States v. Ravara, 2 U.S. (2 Dall.) 297, 298–99 (C.C.D. Pa. 1793) (James Iredell, Richard
Peters, Jr., and James Wilson); Henfield’s Case, 11 F. Cas. 1099, 1100–01 (C.C.D. Pa. 1793) (No.
6360) (John Jay); id. at 1117 (James Wilson).
69 See In re Estate of Ferdinand E. Marcos Human Rights Litig., 978 F.2d 493, 502 (9th Cir.
1992); Filartiga v. Pena-Irala, 630 F.2d 876, 885 (2d Cir. 1980); Xuncax v. Gramajo, 886 F. Supp.
162, 193 (D. Mass. 1995); Forti v. Suarez-Mason, 672 F. Supp. 1531, 1544 (N.D. Cal. 1987); JORDAN J. PAUST, INTERNATIONAL LAW AS LAW OF THE UNITED STATES 7–8 (2d ed. 2003); Jeffrey M. Blum & Ralph G. Steinhardt, Federal Jurisdiction over International Human Rights
Claims: The Alien Tort Claims Act After Filartiga v. Pena-Irala, 22 HARV. INT’L L.J. 53, 57–58
(1981); Edwin D. Dickinson, The Law of Nations as Part of the National Law of the United
States, 101 U. PA. L. REV. 26, 34–46 (1952); Glennon, supra note 10, at 343–47; Koh, supra note 2,
at 1841, 1846; Lobel, supra note 10, at 1090–95; Steven M. Schneebaum, The Enforceability of
Customary Norms of Public International Law, 8 BROOK. J. INT’L L. 289, 289–91 (1982).
70 Koh, supra note 2, at 1841, 1846.
71 630 F.2d 876 (2d Cir. 1980).
884
HARVARD LAW REVIEW
[Vol. 120:869
to approve of the use of the ATS for international human rights litigation) similarly asserted that CIL “has always been part of the federal
common law.”72
By contrast, other courts and scholars, and the Restatement of Foreign Relations Law, concluded that CIL did not have the status of federal law during the pre-Erie period.73 In support of this conclusion,
they cited pre-Erie decisions suggesting that CIL was not federal law
for purposes of Article III or statutory federal question jurisdiction.74
They also noted that the executive branch in the nineteenth century
repeatedly described itself as lacking the authority, in the absence of
congressional authorization, to force the states to comply with CIL.75
Moreover, they pointed out that the most famous general common law
decision — Swift v. Tyson — involved the international law merchant,
a component of the “law of nations” of the time. These scholars contended that phrases like “law of the land,” “law of the United States,”
and “our law” in the nineteenth century were not references to the
“Laws of the United States” in Articles III or VI of the Constitution,76
but rather were phrases commonly used to refer to general common
law.77 Finally, this group argued that CIL’s status as general common
law did not conflict with the Founders’ desire to prevent states from
violating CIL because it merely left the responsibility of policing state
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
72
73
Id. at 885.
See, e.g., Sampson v. Fed. Republic of Germany, 250 F.3d 1145, 1153 n.4 (7th Cir. 2001);
RESTATEMENT (THIRD) OF THE FOREIGN RELATIONS LAW OF THE UNITED STATES ch. 2,
introductory note, at 41 (1987); Bradley & Goldsmith, supra note 3, at 822–26; Clark, supra note
13, at 1283; Stewart Jay, The Status of the Law of Nations in Early American Law, 42 VAND. L.
REV. 819, 832–33 (1989); Arthur M. Weisburd, The Executive Branch and International Law, 41
VAND. L. REV. 1205, 1233–34 (1988); Ernest A. Young, Sorting Out the Debate over Customary
International Law, 42 VA. J. INT’L L. 365, 368 (2002); see also CHARLES PERGLER, JUDICIAL
INTERPRETATION OF INTERNATIONAL LAW IN THE UNITED STATES 19 (1928); QUINCY
WRIGHT, THE CONTROL OF AMERICAN FOREIGN RELATIONS 161 (1922).
74 See, e.g., Ker v. Illinois, 119 U.S. 436, 444 (1886) (holding that the question whether forcible
seizure in a foreign country is grounds to resist trial in state court is “a question of common law,
or of the law of nations” that the Supreme Court has “no right to review”); N.Y. Life Ins. Co. v.
Hendren, 92 U.S. 286, 286–87 (1876) (holding that the Supreme Court has no jurisdiction to review “general laws of war, as recognized by the law of nations applicable to this case,” because
they do not involve “the constitution, laws, treaties, or executive proclamations, of the United
States”); Am. Ins. Co. v. Canter, 26 U.S. (1 Pet.) 511, 545 (1828) (holding that a case involving application of the admiralty and maritime law — elements of the law of nations — “does not . . .
arise under the Constitution or laws of the United States” within the meaning of Article III). In
addition, The Paquete Habana itself strongly suggested the same conclusion when it stated that
CIL as applied by federal courts did not bind either Congress or the President. See The Paquete
Habana, 175 U.S. 677, 700 (1900) (stating that customs and usages of civilized nations govern
“where there is no treaty, and no controlling executive or legislative act”); id. at 708 (stating that
courts must “give effect to” CIL “in the absence of any treaty or other public act of [the] government in relation to the matter”).
75 See Bradley & Goldsmith, supra note 3, at 825 & nn.56–58.
76 U.S. CONST. art. III, § 2, cl. 1; id. art. VI, cl. 2.
77 See Bradley & Goldsmith, supra note 3, at 822–26.
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
885
compliance with international law to the federal political branches,
which could incorporate CIL into federal statutory or treaty law and
could vest federal courts with jurisdiction in cases involving the interpretation of the general common law of CIL.78
B. CIL’s Domestic Status in the Absence
of Political Branch Authorization
The second pre-Sosa debate concerned the effect of Erie on CIL’s
legal status in the United States. In particular, the question was
whether, after Erie, CIL had the status of federal common law.
As we briefly explained in the Introduction, modern position proponents maintained that, after Erie, CIL had the status of selfexecuting federal common law that federal courts were bound to
apply even in the absence of political branch authorization.79 This
claim rested in part on the historical proposition, outlined earlier, that
CIL had the status of federal law since the Founding and thus remained unaffected by Erie’s rejection of federal general common law.
It also relied on the Supreme Court’s holding in Sabbatino that the act
of state doctrine, while not required by CIL, is based on principles of
separation of powers and is therefore a rule of federal common law
binding on the states.80 The Court in Sabbatino reasoned that the act
of state doctrine should be subject to a uniform national standard because it was “concerned with a basic choice regarding the competence
and function of the Judiciary and the National Executive in ordering
our relationships with other members of the international community.”81 Supporters of the modern position argued that CIL similarly
concerns the United States’s relationship with the international
community.82
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
78
79
See id. at 871.
See, e.g., Filartiga v. Pena-Irala, 630 F.2d 876, 887 n.20 (2d Cir. 1980) (expressing the view
that CIL “has an existence in the federal courts independent of acts of Congress”); RESTATEMENT (THIRD) OF THE FOREIGN RELATIONS LAW OF THE UNITED STATES § 111(3) (1987)
(“Courts in the United States are bound to give effect to [customary] international law . . . .”); id.
§ 115 cmt. e (“[A]ny rule of customary international law . . . is federal law . . . [and] supersedes
inconsistent State law or policy whether adopted earlier or later.”); Brilmayer, supra note 2, at 324
(asserting that “whatever [customary] international law requires, it is binding on the states” (emphasis omitted)); Henkin, supra note 2, at 1561 (stating that CIL “is ‘self-executing’ and is applied
by courts in the United States without any need for it to be enacted or implemented by
Congress”).
80 See Banco Nacional de Cuba v. Sabbatino, 376 U.S. 398, 425–27 (1964). The Court in Sabbatino said that this conclusion was similar to the position taken by Professor Philip Jessup in an
article arguing that, despite Erie, state courts should not have the final word on the interpretation
of international law. See id. at 425 (citing Philip C. Jessup, The Doctrine of Erie Railroad v.
Tompkins Applied to International Law, 33 AM. J. INT’L L. 740 (1939)).
81 Id. at 425.
82 See, e.g., Henkin, supra note 2; Alfred Hill, The Law-Making Power of the Federal Courts:
Constitutional Preemption, 67 COLUM. L. REV. 1024, 1057–68 (1967).
886
HARVARD LAW REVIEW
[Vol. 120:869
By contrast to the modern position, the revisionist view was that
CIL does not automatically have the status of federal common law and
that after Erie, federal courts needed some authorization from either
the political branches or the Constitution in order to apply CIL.83 The
revisionist view built on the historical contention that CIL was not
federal law prior to Erie. It argued that, after Erie, neither the Constitution nor any federal statute authorizes the modern position’s envisioned wholesale application of CIL by the federal judiciary. Articles
III and VI of the Constitution both refer to treaties, but not CIL, in
their list of federal laws, and the Constitution’s only reference to CIL
is in the Article I Define and Punish Clause.84 Revisionists argued
that the constitutional text therefore suggests that Congress must act
before CIL is incorporated into domestic law. As for Sabbatino, revisionists noted that the act of state doctrine, as articulated in Sabbatino,
was designed to prevent judicial involvement in foreign affairs and
was grounded in principles of separation of powers. They then argued
that the primary application of the modern position concerns human
rights cases against foreign governments — cases that, contrary to Sabbatino’s central premise, place federal courts in the center of foreign
affairs controversies. They also noted that the application of a CIL of
human rights as federal common law would be contrary to the postErie requirement that federal common law conform to the policies of
the federal political branches. Congress has incorporated only select
CIL principles into federal statutory law, and in the human rights context in which the modern position matters most, said revisionists, the
political branches have made clear through their limitations on U.S.
ratification of human rights treaties that they do not want international human rights norms to provide a basis for domestic litigation.
An intermediate position that emerged prior to Sosa was that federal courts could apply CIL as a type of law that was neither state law
nor preemptive federal law.85 Under the main variant of this approach, CIL would be applied like pre-Erie general common law and
thus would be available as a rule of decision for federal and state
courts but would neither preempt state law nor provide a basis for
federal question jurisdiction.86 Under a different variant, CIL would
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
83
84
See sources cited supra note 3.
See U.S. CONST. art. I, § 8, cl. 10 (granting Congress the power “[t]o define and punish . . .
Offences against the Law of Nations”).
85 See T. Alexander Aleinikoff, International Law, Sovereignty, and American Constitutionalism: Reflections on the Customary International Law Debate, 98 AM. J. INT’L L. 91 (2004); Michael D. Ramsey, International Law as Non-Preemptive Federal Law, 42 VA. J. INT’L L. 555
(2002); Weisburd, supra note 3; Young, supra note 73. It is possible that this is what Professor
Jessup had in mind. See Jessup, supra note 80.
86 See Aleinikoff, supra note 85, at 97; Weisburd, supra note 3, at 48–49; Young, supra note 73,
at 467–68.
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
887
be treated as federal law for purposes of jurisdiction under Article III
of the Constitution, but not for purposes of preemption under the Constitution’s Supremacy Clause.87
C. Did the ATS Authorize Courts To Apply CIL as Federal Law?
The third debate prior to Sosa concerned whether there was, consistent with the usual requirements for post-Erie federal common law,
any congressional statute that authorized federal courts to apply CIL
as federal law. This debate was distinct from, though potentially related to, the modern position debate. One could reject the view that
CIL is self-executing federal common law and believe nonetheless that
Congress has authorized federal courts to apply CIL as federal law in
certain cases.88 The main focus of this debate was the ATS, the fount
of modern international human rights litigation. Is the ATS a mere jurisdictional statute, or does it also create a cause of action for human
rights abuses or otherwise authorize courts to apply CIL in cases properly brought under the ATS? A variety of answers were offered.
One answer was that the ATS is a purely jurisdictional statute that
authorizes nothing with regard to substantive law.89 This view rested
primarily on the plain language of the ATS. Enacted as part of the
Judiciary Act of 178990 — a statute that regulated the jurisdiction and
structure of the federal courts, not causes of action — the ATS’s original language provided that federal district courts “shall also have cognizance . . . of all causes where an alien sues for a tort only in violation
of the law of nations or a treaty of the United States.”91 “Cognizance”
was a term of art referring to jurisdiction, and the First Congress used
different language when it created statutory civil actions.92 The jurisdictional reading of the ATS found support in the current codification
of the statute, which extends “original jurisdiction” over certain cases
brought by aliens and does not refer to damages or other remedies.93
This jurisdictional reading of the ATS was consistent with the decision that reinvigorated the ATS in modern times, Filartiga v. PenaIrala. The Second Circuit in Filartiga construed the ATS, for purposes of its decision, “not as granting new rights to aliens, but simply
as opening the federal courts for adjudication of the rights already rec–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
87
88
89
See Ramsey, supra note 85, at 575–77.
See Bradley & Goldsmith, supra note 3, at 872–73 & nn.352–54.
See, e.g., Curtis A. Bradley, The Alien Tort Statute and Article III, 42 VA. J. INT’L L. 587,
591 (2002); Curtis A. Bradley & Jack L. Goldsmith III, The Current Illegitimacy of International
Human Rights Litigation, 66 FORDHAM L. REV. 319, 358 (1997).
90 ch. 20, 1 Stat. 73.
91 Id. § 9(b), 1 Stat. at 77.
92 See William R. Casto, The Federal Courts’ Protective Jurisdiction over Torts Committed in
Violation of the Law of Nations, 18 CONN. L. REV. 467, 479 (1986).
93 28 U.S.C. § 1350 (2000).
888
HARVARD LAW REVIEW
[Vol. 120:869
ognized by international law.”94 The court proceeded to hold that a
human rights suit brought by Paraguayan plaintiffs against a former
Paraguayan official satisfied Article III because CIL was part of federal common law and the plaintiffs’ CIL claim therefore arose under
federal law.95 The court in Filartiga did not hold, however, that either
CIL itself or the ATS created the plaintiffs’ cause of action. The court
insisted that “the question of federal jurisdiction under the Alien Tort
Statute, which requires consideration of the law of nations,” and the
“issue of the choice of law to be applied” were “distinct” issues and remanded to the district court to determine whether Paraguayan law or
some other law governed the merits of the suit.96
A number of courts after Filartiga, however, including the Second
Circuit, held that the ATS both established federal jurisdiction and
also created a substantive federal cause of action for torts in violation
of CIL.97 The most prominent argument in support of this position
was that the 1992 Torture Victim Protection Act98 (TVPA), particularly
its legislative history, confirmed that Congress had authorized causes
of action in ATS litigation.99 At least one court adopted a somewhat
different position, interpreting the ATS not as creating a cause of action, but rather as authorizing the federal courts to do so.100 The
analogy for this latter position was the Supreme Court’s decision in
Textile Workers Union v. Lincoln Mills,101 which implied federal common law–making powers from the Labor Management Relations Act’s
grant of federal jurisdiction to decide disputes under certain labormanagement contracts.102
D. Scope and Sources of CIL in ATS Litigation
The ATS refers to torts in violation of the “law of nations,” but it
does not specify which law of nations rules can be applied or how to
discern their content. Prior to Sosa, there was disagreement among
courts and scholars over the scope and sources of CIL in ATS litiga–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
94
95
96
97
Filartiga v. Pena-Irala, 630 F.2d 876, 887 (2d Cir. 1980).
Id. at 886–87.
Id. at 889.
See, e.g., Kadic v. Karadžic, 70 F.3d 232, 241 (2d Cir. 1995); In re Estate of Ferdinand Marcos, Human Rights Litig., 25 F.3d 1467, 1475 (9th Cir. 1994).
98 Pub. L. No. 102-256, 106 Stat. 73 (1992) (codified at 28 U.S.C. § 1350 note).
99 See, e.g., Kadic v. Karadzic, 74 F.3d 377, 378 (2d Cir. 1996) (stating, in denying rehearing of
its earlier decision, that “Congress ha[d] made clear that its enactment of the Torture Victim Protection Act . . . was intended to codify the cause of action recognized by this Circuit in Filártiga,
even as it extends the cause of action to plaintiffs who are United States citizens”). The TVPA
provides a federal cause of action for acts of torture and “extrajudicial killing” committed under
authority of foreign law. See 28 U.S.C. § 1350 note.
100 See Abebe-Jira v. Negewo, 72 F.3d 844, 848 (11th Cir. 1996).
101 353 U.S. 448 (1957).
102 See id. at 456–57.
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
889
tion. In terms of scope, the issue was whether plaintiffs could bring
claims under all CIL norms relating to torts or only under a subset of
such norms. In terms of sources, the issue was whether, in discerning
the content of CIL, courts should focus on verbal evidence of state positions such as treaties and U.N. General Assembly resolutions or
should instead focus primarily on state practice. Like the other debates discussed earlier in this Article, this debate over the scope and
sources of CIL occurred against the backdrop of the narrow role for
judicial lawmaking envisioned by Erie.
On the scope issue, the court in Filartiga suggested that all CIL
rules relating to torts were eligible for ATS litigation. For guidance on
the proper standard for determining rules of CIL, the court relied on
pre-Erie decisions applying CIL outside the context of the ATS, such
as The Paquete Habana. Based on these decisions, the court stated
that, in order for a norm to qualify as a rule of CIL, it must “command
the ‘general assent of civilized nations.’”103 Although the court described this requirement as “stringent,” it viewed it as a general requirement for CIL, not a requirement unique to the ATS context.104
The court also concluded that it should interpret international law not
as it was in 1789 when the ATS was enacted, but “as it has evolved
and exists among the nations of the world today,”105 again suggesting
that all rules properly found to be CIL would qualify.
On the sources issue, the court in Filartiga relied primarily on verbal evidence of state assent rather than on state practice. In concluding that official torture violated CIL, the court relied on, for example:
references to human rights in the United Nations Charter; the U.N.
General Assembly’s nonbinding Universal Declaration of Human
Rights;106 another nonbinding General Assembly resolution concerning
torture; various treaties that the United States had not at that time
ratified; and the prohibitions on torture in a number of national constitutions.107 The court did not maintain that the international community had abolished torture in practice. Indeed, the court acknowledged
that the prohibition on torture is “often honored in the breach,” but
observed that it was not aware of any nation that verbally asserted
that it had the right to engage in torture.108
After Filartiga, some courts sought to limit the scope of CIL claims
by suggesting that only certain well-defined and widely accepted CIL
norms could be brought in ATS litigation. The Ninth Circuit, for ex–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
103
104
105
106
107
108
Filartiga v. Pena-Irala, 630 F.2d 876, 881 (2d Cir. 1980).
Id.
Id.
G.A. Res. 217A (III), U.N. GAOR, 3d Sess., U.N. Doc. A/810 (Dec. 12, 1948).
See Filartiga, 630 F.2d at 881–84.
Id. at 884 & n.15.
890
HARVARD LAW REVIEW
[Vol. 120:869
ample, stated that “[a]ctionable violations of international law must be
of a norm that is specific, universal, and obligatory.”109 Other courts
suggested that the ATS is limited to particularly egregious violations of
CIL. The Second Circuit stated, for example, that the ATS “applies
only to shockingly egregious violations of universally recognized principles of international law.”110 Some litigants and commentators suggested that ATS litigation should be limited to violations of jus cogens
norms.111 A jus cogens norm is a norm “accepted and recognized by
the international community of States as a whole as a norm from
which no derogation is permitted and which can be modified only by a
subsequent norm of general international law having the same character.”112 In the Ninth Circuit opinion that the Supreme Court reviewed
in Sosa, the court rejected such a jus cogens limitation, explaining:
The notion of jus cogens norms was not part of the legal landscape
when Congress enacted the [ATS] in 1789. Thus, to restrict actionable
violations of international law to only those claims that fall within the
categorical universe known as jus cogens would deviate from both the history and text of the [ATS].113
With respect to the sources of CIL, the Second Circuit eventually
pulled back from the approach in Filartiga, which, as we noted earlier,
had relied heavily on verbal statements and “consensus” and had
downplayed actual practice. In particular, in Flores v. Southern Peru
Copper Corp.,114 the court held that, “[i]n determining whether a particular rule is a part of customary international law — i.e., whether
States universally abide by, or accede to, that rule out of a sense of legal obligation and mutual concern — courts must look to concrete evidence of the customs and practices of States.”115 In other words, just
as with the issue of scope, some courts began to develop a revisionist
position with respect to the sources of CIL in ATS litigation. Some
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
109
110
In re Estate of Ferdinand Marcos, Human Rights Litig., 25 F.3d 1467, 1475 (9th Cir. 1994).
Zapata v. Quinn, 707 F.2d 691, 692 (2d Cir. 1983) (per curiam); see also Beanal v. FreeportMcMoran, Inc., 197 F.3d 161, 167 (5th Cir. 1999). The Second Circuit later clarified that “Zapata
does not establish ‘shockingly egregious’ as an independent standard for determining whether
alleged conduct constitutes a violation of international law.” Flores v. S. Peru Copper Corp., 414
F.3d 233, 253 (2d Cir. 2003).
111 See, e.g., Ryan Goodman & Derek P. Jinks, Filartiga’s Firm Footing: International Human
Rights and Federal Common Law, 66 FORDHAM L. REV. 463, 495 (1997).
112 Vienna Convention on the Law of Treaties art. 53, opened for signature May 23, 1969, 1155
U.N.T.S. 331.
113 Alvarez-Machain v. United States, 331 F.3d 604, 614 (9th Cir. 2003) (citation omitted), rev’d
on other grounds sub nom. Sosa v. Alvarez-Machain, 124 S. Ct. 2739 (2004).
114 414 F.3d 233 (2d Cir. 2003).
115 Id. at 250 (emphasis omitted); see also United States v. Yousef, 327 F.3d 56, 103 & n.37 (2d
Cir. 2003) (favoring “formal lawmaking and official actions of States” over scholarly opinions as
proper bases for determining states’ practices). In Flores, the court rejected the claim that intranational pollution violates CIL. See Flores, 414 F.3d at 255.
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
891
scholars, too, began to criticize reliance on verbal statements and “consensus” as a basis for CIL, and to argue for a renewed emphasis on
state practice.116
Table 1 illustrates the four pre-Sosa debates described above:
TABLE 1
Pre-Erie Status
of CIL
Post-Erie Status
of CIL
Nature of ATS
Scope and Sources of
CIL to be Applied
by Courts in ATS
Litigation
Conventional
Wisdom in 1990s
Federal law
Revisionist View
Wholesale
incorporation as
federal common law
Selective incorporation
based on constitutional
or political branch
authorization
Either creates federal
causes of action or
authorizes courts to create them
All of CIL, derived from
wide range of materials
Only jurisdictional
General law
Limited set of CIL norms,
based primarily on the
practice of nations
IV. SOSA, THE ATS, AND THE MODERN POSITION
In this Part, we analyze Sosa’s implications for the four debates
discussed above. As we demonstrate, Sosa directly resolved two of the
four debates. With respect to the first debate over the pre-Erie status
of CIL, the Court clearly understood that CIL historically had the
status of nonfederal general common law. With respect to the third
debate over whether the ATS authorizes the federal courts to create
common law causes of action based on CIL, the Court concluded that
even though the ATS was originally intended as only a jurisdictional
statute, Congress’s expectations in enacting the ATS have the effect
today of authorizing courts to develop a narrow set of federal common
law causes of action. In addition to resolving these debates, the Court
in Sosa strongly suggested that, with respect to the fourth debate over
the scope and sources of CIL to be applied in ATS litigation, courts
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
116 See, e.g., J. Patrick Kelly, The Twilight of Customary International Law, 40 VA. J. INT’L L.
449 (2000).
892
HARVARD LAW REVIEW
[Vol. 120:869
should discern and apply CIL more carefully and cautiously than
many lower courts did prior to Sosa. Finally, although the Court in
Sosa did not specifically address the second debate over whether CIL
is automatically part of the post-Erie federal common law, the Court’s
reasoning and conclusions are incompatible with the claim that CIL
has this status. We assess Sosa’s implications for this second debate
last because these implications are the most complex and are related to
the ways in which Sosa resolved the other debates.
A. CIL’s Pre-Erie Status
In Sosa, the Supreme Court dismissed Alvarez-Machain’s CIL
claim on the basis of a complicated chain of reasoning. A critical link
in that chain concerned the pre-Erie status of CIL. Although the
Court acknowledged that U.S. courts had applied CIL since the
Founding, it made clear that before Erie the CIL they applied had the
status of general common law, not federal common law. This conclusion, in turn, led the Court to consider carefully the implications of
Erie for the domestic status of CIL.
The Court repeatedly described the few law of nations claims that
it thought could have been brought historically under the ATS as part
of the pre-Erie “common law.”117 The Court also invoked two famous
Holmesian descriptions of general common law in the context of referring to the pre-Erie domestic status of the law of nations.118 Relatedly,
the Court explained that the law of nations in the nineteenth century
encompassed subjects such as the international “law merchant”119 that
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
117 See Sosa v. Alvarez-Machain, 124 S. Ct. 2739, 2754 (2004) (“[A]t the time of enactment the
jurisdiction [conferred by the ATS] enabled federal courts to hear claims in a very limited category defined by the law of nations and recognized at common law.”); id. at 2759 (“[S]ome, but few,
torts in violation of the law of nations were understood to be within the common law.”); id.
(“[T]he ATS was meant to underwrite litigation of a narrow set of common law actions derived
from the law of nations . . . .”); id. at 2764 (“[T]he jurisdiction [conferred by the ATS] was originally understood to be available to enforce a small number of international norms that a federal
court could properly recognize as within the common law enforceable without further statutory
authority.”).
118 Id. at 2762 (“[T]he prevailing conception of the common law has changed since 1789 in a
way that counsels restraint in judicially applying internationally generated norms. When § 1350
was enacted, the accepted conception was of the common law as ‘a transcendental body of law
outside of any particular State but obligatory within it unless and until changed by statute.’”
(quoting Black & White Taxicab & Transfer Co. v. Brown & Yellow Taxicab & Transfer Co., 276
U.S. 518, 533 (1928) (Holmes, J., dissenting))); id. at 2760 (noting that the argument that the Continental Congress would have had no reason to recommend that States enact statutes to duplicate
international law remedies already available at common law “rests on a misunderstanding of the
relationship between common law and positive law in the late 18th century, when positive law
was frequently relied upon to reinforce and give standard expression to the ‘brooding omnipresence’ of the common law then thought discoverable by reason” (footnote omitted) (quoting S. Pac.
Co. v. Jensen, 244 U.S. 205, 222 (1917) (Holmes, J., dissenting))).
119 Id. at 2756.
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
893
were indisputably part of the pre-Erie general common law.120 The
Court then stated that “it was the law of nations in this sense” — the
same general common law sense as the law merchant — “that our precursors spoke about when the Court explained [in The Paquete Habana] the status of coast fishing vessels in wartime grew from ‘ancient
usage among civilized nations, beginning centuries ago, and gradually
ripening into a rule of international law.’”121 And the Court noted the
change in the prevailing conception of the “common law” brought
about in Erie, thus further linking the pre-Erie status of the law of nations with general common law.122
These passages show that the Court in Sosa understood CIL as
having general law, rather than federal law, status prior to Erie. This
conclusion is reinforced by the Court’s extensive consideration of what
it referred to as the “watershed” decision in Erie and its implications
— consideration that would have been largely unnecessary if CIL were
federal law, rather than general law, prior to Erie. As a result, Sosa
repudiated a central historical claim made by many proponents of the
modern position — that is, that CIL historically had the status of federal law and thus lay outside of Erie’s reach.
B. Sosa and the ATS
Sosa, and the U.S. government acting as amicus curiae, argued that
the ATS was simply a jurisdictional statute that did not create a cause
of action for violations of CIL. Alvarez-Machain, by contrast, argued
that the ATS did create CIL causes of action, including a cause of action for arbitrary arrest.
The Court unanimously concluded that “the ATS is a jurisdictional
statute creating no new causes of action.”123 The Court reasoned that
the original ATS provided that courts would have “cognizance” of certain causes of action, a term that referred to jurisdiction.124 It further
noted that the ATS was placed in § 9 of the Judiciary Act of 1789, “a
statute otherwise exclusively concerned with federal-court jurisdiction.”125 In his concurrence, Justice Scalia (joined by Chief Justice
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
120
121
122
See Swift v. Tyson, 41 U.S. (16 Pet.) 1 (1842).
Sosa, 124 S. Ct. at 2756 (quoting The Paquete Habana, 175 U.S. 677, 686 (1900)).
See id. at 2762. In addition, Justice Scalia’s concurrence explained at length that CIL had
the status of general common law before Erie. Id. at 2769–71 (Scalia, J., concurring). The majority opinion disputed several aspects of Justice Scalia’s concurrence, but not this one.
123 Id. at 2761 (majority opinion); see also id. at 2754 (“[W]e agree the statute is in terms only
jurisdictional . . . .”); id. at 2755 (referring to the ATS’s “strictly jurisdictional nature”); id. (“[W]e
think the statute was intended as jurisdictional in the sense of addressing the power of the courts
to entertain cases concerned with a certain subject.”).
124 See id. at 2755.
125 Id.
894
HARVARD LAW REVIEW
[Vol. 120:869
Rehnquist and Justice Thomas) agreed with this analysis.126 Thus, as
the Court explained, “[a]ll Members of this Court agree that § 1350 is
only jurisdictional.”127
The Court in Sosa believed that its holding that the ATS is only a
jurisdictional statute “raise[d] a new question . . . about the interaction
between the ATS at the time of its enactment and the ambient law of
the era.”128 Exploration of this new question led the Court to conclude
that, although the ATS was not intended to create causes of action related to CIL, it has the effect today of authorizing federal courts to
recognize post-Erie federal common law causes of action for a limited
number of CIL violations. The Court reached this conclusion in three
steps.
First, the Court reasoned that the Congress that enacted the ATS in
1789 assumed that there would be preexisting law, with the status of
general common law, to apply in cases within ATS jurisdiction. As the
Court explained, “federal courts could entertain claims once the jurisdictional grant was on the books, because torts in violation of the law
of nations would have been recognized within the common law of the
time.”129 The Court rejected Sosa’s argument (and the executive
branch’s argument as amicus curiae) that the 1789 Congress believed
that the law of nations component of ATS jurisdiction would lie fallow
unless and until Congress separately enacted statutory causes of action
to be applied in ATS cases.130 Rather, the historical materials persuaded the Court that “the statute was intended to have practical effect the moment it became law”131 and that Congress thought the prac-
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
126
127
See id. at 2772 (Scalia, J., concurring).
Id. at 2764 (majority opinion). It is also worth noting that all nine Justices in Sosa referred
to 28 U.S.C. § 1350 as the “Alien Tort Statute,” not the “Alien Tort Claims Act,” despite disagreement in the briefs over the proper title for the statute. Compare Brief of Petitioner at i, Sosa, 124
S. Ct. 2739 (No. 03-339) (referring to the “Alien Tort Statute” in identifying the questions presented), with Brief for the Respondent at i, Sosa, 124 S. Ct. 2739 (No. 03-485) (referring to the
“Alien Tort Claims Act” in identifying the questions presented). The latter title had been favored
by those advocating a cause-of-action construction of the ATS. See Bradley, supra note 89, at
592–93.
128 Sosa, 124 S. Ct. at 2755.
129 Id.
130 See id. at 2758 (“[T]here is every reason to suppose that the First Congress did not pass the
ATS as a jurisdictional convenience to be placed on the shelf for use by a future Congress or state
legislature that might, some day, authorize the creation of causes of action or itself decide to make
some element of the law of nations actionable for the benefit of foreigners.”); id. at 2758–59
(“There is too much in the historical record to believe that Congress would have enacted the ATS
only to leave it lying fallow indefinitely.”).
131 Id. at 2761.
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
895
tical effect would be guaranteed by preexisting CIL-related causes of
action available at common law.132
Second, the Court concluded that it should preserve the 1789 Congress’s background expectation that there would be common law
causes of action available for judicial application under the ATS. The
Court acknowledged that the common law applied in 1789 differed
significantly from the common law that federal courts applied after
Erie. But the Court thought “it would be unreasonable to assume that
the First Congress would have expected federal courts to lose all capacity to recognize enforceable international norms simply because the
common law might lose some metaphysical cachet on the road to modern realism.”133 In effect, the Court attempted to “translate” the First
Congress’s expectations about the effect of the ATS, which rested on a
pre-Erie understanding of general common law, into the contemporary
context in which federal courts apply nonstate common law only in
specialized circumstances.134 In justifying this conclusion, the Court
noted that “no development in the two centuries from the enactment of
§ 1350 to the birth of the modern line of [ATS] cases . . . has categorically precluded federal courts from recognizing a claim under the law
of nations as an element of common law.”135 In particular, neither
Erie nor Congress had categorically prohibited the judicial recognition
of claims under CIL.136
Third, the Court concluded that the best translation of the original
ATS in the post-Erie world is that the ATS authorizes the judicial
creation of a domestic remedy, in the form of a cause of action, for a
narrow set of CIL violations.137 Thus, as Justice Scalia’s concurrence
explained (and criticized), the Court inferred, from a jurisdictional
statute that enabled courts to apply CIL as general common law, the
authorization for courts to create causes of action for CIL violations, in
narrow circumstances, as a matter of post-Erie federal common law.138
The Court did not explain how this conclusion was consistent with its
description of the ATS as only jurisdictional or with its view that “[t]he
vesting of jurisdiction in the federal courts does not in and of itself
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
132 See id. (“The jurisdictional grant is best read as having been enacted on the understanding
that the common law would provide a cause of action for the modest number of international law
violations with a potential for personal liability at the time.”).
133 Id. at 2765.
134 For discussion of a similar idea of translation in the constitutional context, compare Lawrence Lessig, Fidelity in Translation, 71 TEX. L. REV. 1165 (1993), with Michael J. Klarman, Antifidelity, 70 S. CAL. L. REV. 381 (1997). See also Lawrence Lessig, Erie-Effects of Volume 110:
An Essay on Context in Interpretive Theory, 110 HARV. L. REV. 1785 (1997).
135 Sosa, 124 S. Ct. at 2761.
136 See id. at 2761–62.
137 See id. at 2761–65.
138 See id. at 2773 n.* (Scalia, J., concurring).
896
HARVARD LAW REVIEW
[Vol. 120:869
give rise to authority to formulate federal common law.”139 But its description of the legitimate bases of post-Erie federal common law included a citation to Textile Workers Union v. Lincoln Mills,140 a decision in which (as we noted earlier) the Court implied federal common
law–making powers from the Labor Management Relations Act’s
grant of federal jurisdiction to decide disputes under certain labormanagement contracts.141 Presumably, the common law powers recognized in Sosa were similar.142
We should make clear that we are merely describing the Court’s
reasoning here, not defending it. The idea of “translating” law to fit
with changed circumstances is controversial even in constitutional law,
where formal amendment of the law is much more difficult than with
a statute like the ATS. Moreover, even assuming translation is sometimes appropriate with respect to statutes, it can reasonably be argued
that the substantial changes in both the nature of the common law and
the content of international law make translating the “law of nations”
prong of the ATS too difficult. (This was, essentially, Justice Scalia’s
position.) There is also tension, if not outright contradiction, in the
Court’s construction of the ATS as both purely jurisdictional and an
authorization for creating causes of action. Full exploration of this potential problem in the Court’s analysis is beyond the scope of this project. The key point for present purposes is simply that the Court
based its allowance of CIL claims under the ATS on its understanding
of Congress’s intent in enacting the ATS.
C. Scope and Sources of CIL in ATS Litigation
The Court in Sosa limited its holding that the ATS authorizes federal courts to recognize federal common law causes of action based on
CIL by requiring that any such recognition satisfy at least two requirements. First, the CIL norm in question must be “accepted by the
civilized world” to the same degree as the few law of nations norms
that the First Congress would have expected to be enforceable through
private claims in 1789.143 Second, the CIL norm in question must be
“defined with a specificity” comparable to the historic law of nations
norms.144 The Court added that the evaluation of “whether a norm is
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
139
140
141
142
Tex. Indus., Inc. v. Radcliff Materials, Inc., 451 U.S. 630, 640–41 (1981).
353 U.S. 448 (1957).
See id. at 456–57.
For criticism of the Court’s reasoning on this issue, see Note, An Objection to Sosa — And
to the New Federal Common Law, 119 HARV. L. REV. 2077 (2006).
143 Sosa, 124 S. Ct. at 2761. These norms involved violations of safe conduct, infringement of
the rights of ambassadors, and piracy. For discussion of the right of safe conduct, see Thomas H.
Lee, The Safe-Conduct Theory of the Alien Tort Statute, 106 COLUM. L. REV. 830 (2006).
144 Sosa, 124 S. Ct. at 2761.
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
897
sufficiently definite to support a cause of action should (and, indeed,
inevitably must) involve an element of judgment about the practical
consequences of making that cause available to litigants in the federal
courts.”145
The Court made clear that this test for domestic enforcement of
CIL under the ATS is more demanding than the test for whether a
CIL norm is internationally binding according to the traditional standards for CIL.146 It made clear, in other words, that the CIL violations Congress made actionable in ATS cases are a subset of all CIL
violations. Applying this two-part test, the Court concluded that Alvarez-Machain did not allege a violation of a norm of CIL so well defined and accepted as to support the creation of a federal cause of action.147 Although this conclusion is easy to state, the Court’s analysis
of the sources and scope of the CIL available in ATS cases raises a
number of questions. In what follows, we explore these questions, and
explain why, despite ambiguities in some places, the opinion is best
read as significantly limiting the causes of action available in ATS
cases.
Consider first the Court’s “clear definition” requirement. Many
lower courts prior to Sosa had not required a close correspondence between the content of the CIL sources relied on by plaintiffs and their
causes of action. The Supreme Court in Sosa took a stricter approach.
For example, the Court maintained that the recognition by national
constitutions of a prohibition on arbitrary detention reflected a consensus at too “high [a] level of generality” to support Alvarez-Machain’s
claim for relief for a one-day detention not authorized by law.148 Similarly, it found insufficient the Restatement of Foreign Relations Law’s
claim that a “state policy”149 of “prolonged arbitrary detention”150 was
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
145
146
Id. at 2766 (footnote omitted).
See id. at 2768 (“Alvarez cites little authority that a rule so broad has the status of a binding
customary norm today. He certainly cites nothing to justify the federal courts in taking his broad
rule as the predicate for a federal lawsuit . . . .” (footnote omitted)); id. at 2769 (“Even the Restatement’s limits [on the CIL rule concerning arbitrary detention] are only the beginning of the
enquiry, because although it is easy to say that some policies of prolonged arbitrary detentions are
so bad that those who enforce them become enemies of the human race, it may be harder to say
which policies cross that line with the certainty afforded by Blackstone’s three common law offenses.”); id. at 2769 n.29 (“[T]hat a rule as stated is as far from full realization as the one Alvarez
urges is evidence against its status as binding law; and an even clearer point against the creation
by judges of a private cause of action to enforce the aspiration behind the rule claimed.”); cf.
Daniel J. Meltzer, Customary International Law, Foreign Affairs, and Federal Common Law, 42
VA. J. INT’L L. 513, 519 (2002) (noting that “the fact that a rule has been recognized as CIL, by
itself, is not an adequate basis for viewing that rule as part of federal common law”).
147 See Sosa, 124 S. Ct. at 2766–69.
148 Id. at 2768 n.27.
149 Id. at 2768 (quoting RESTATEMENT (THIRD) OF THE FOREIGN RELATIONS LAW OF
THE UNITED STATES § 702 (1987)).
898
HARVARD LAW REVIEW
[Vol. 120:869
a violation of CIL, in part because the Restatement required “a factual
basis beyond relatively brief detention in excess of positive authority.”151 The Sosa decision thus seems to limit causes of action in ATS
cases to those for which the content of the CIL norm corresponds
closely with the plaintiff’s sources.152 This conclusion is consistent
with the Court’s insistence, in another part of the opinion, that it had
“no congressional mandate to seek out and define new and debatable
violations of the law of nations.”153
The meaning of the “acceptance” prong of the Sosa test for recognizing a cause of action in ATS cases is less certain. As we noted earlier, many pre-Sosa lower court decisions downplayed the traditional
state practice requirement for CIL and emphasized instead state acceptance as reflected in instruments like General Assembly resolutions,
multilateral treaties, national constitutions, and official pronouncements of international bodies.154 Sosa appears to render some of these
sources irrelevant, minimize the significance of others, and reemphasize the importance of looking to state practice in ATS cases.
The Court in Sosa first looked to the Universal Declaration of
Human Rights, a U.N. General Assembly resolution outlining fundamental human rights norms that pre-Sosa courts had relied on heavily
in identifying causes of action in ATS cases.155 The Court declined to
rely on this source as a basis for a CIL cause of action, noting correctly
that the “Declaration does not of its own force impose obligations as a
matter of international law” and concluding that the Declaration did
not itself “establish the relevant and applicable rule of international
law.”156 Although the Court went on to acknowledge that the Declaration had a “substantial indirect effect on international law,”157 it also
noted that the Declaration had “little utility under the standard set out
in this opinion,” and the Court did not consider the Declaration further
in its analysis of whether Alvarez-Machain’s proposed norm of arbitrary detention had become so well accepted as to warrant a cause of
action in ATS cases.158
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
150 Id. (emphasis added) (quoting RESTATEMENT (THIRD) OF THE FOREIGN RELATIONS
LAW OF THE UNITED STATES § 702).
151 Id. at 2769.
152 See id. at 2768.
153 Id. at 2763.
154 See supra section III.D, pp. 888–91.
155 See Sosa, 124 S. Ct. at 2767; see also Universal Declaration of Human Rights, supra note
106, at 71.
156 Sosa, 124 S. Ct. at 2767.
157 Id. at 2767 n.23.
158 Id. at 2767.
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
899
The Court reached a similar conclusion with respect to the International Covenant on Civil and Political Rights159 (ICCPR). Like the
Universal Declaration, the ICCPR was widely relied upon in pre-Sosa
ATS cases for the identification of CIL causes of action. The Court in
Sosa noted that the ICCPR, unlike the Universal Declaration, was a
ratified treaty and therefore bound the United States as a matter of international law. But the Court added that the ICCPR was “not selfexecuting and so did not itself create obligations enforceable in the
federal courts.”160 For this reason, the Court concluded that the
ICCPR, like the Universal Declaration, had “little utility” under the
Sosa standard for identifying CIL causes of action.161 Although the
Court did mention the ICCPR in its subsequent analysis of CIL, it did
so only in a negative way. After describing Alvarez-Machain’s claim
that the CIL prohibition on “arbitrary detention” extended to any brief
detention not sanctioned by domestic law, the Court added that
“[w]hether or not this is an accurate reading of the [ICCPR], Alvarez
cites little authority that a rule so broad has the status of a binding
customary norm today.”162 The clear implication is that, contrary to
lower court practice prior to Sosa,163 the presence of a norm in the
ICCPR no longer provides significant evidence of a CIL cause of action in ATS cases.164
The Court in Sosa also considered national constitutions, which
showed that “many nations recognize a norm against arbitrary detention.”165 This too was a source that lower courts had considered prior
to Sosa.166 In contrast to the Universal Declaration and the ICCPR,
the Court implied that this source might be influential in establishing
that nations had accepted the norm in question.167 But as explained
earlier, the Court dismissed this source because the consensus against
arbitrary detention reflected in the constitutions was at a significantly
higher level of generality than Alvarez-Machain’s claim.168
Yet another source the Court considered was the Restatement of
Foreign Relations Law. Once again, lower courts had relied heavily
and uncritically on the Restatement in developing federal common law
causes of action under the ATS. A number of courts viewed the Re–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
159
160
161
162
163
164
Dec. 19, 1966, 999 U.N.T.S. 171, 6 I.L.M. 368.
Sosa, 124 S. Ct. at 2767.
Id.
Id. at 2768.
See, e.g., Alvarez-Machain v. United States, 331 F.3d 604, 620–21 (9th Cir. 2003).
See, e.g., Aldana v. Del Monte Fresh Produce, N.A., Inc., 416 F.3d 1242, 1247 (11th Cir.
2005) (disapproving pre-Sosa district court decisions that had relied on the ICCPR).
165 Sosa, 124 S. Ct. at 2768 n.27.
166 See, e.g., Alvarez-Machain, 331 F.3d at 620.
167 See Sosa, 124 S. Ct. at 2768 n.27.
168 See id. at 2768.
900
HARVARD LAW REVIEW
[Vol. 120:869
statement’s list of customary international human rights norms as actionable under the ATS.169 The Court in Sosa rejected this approach.
It explained that whether a norm was included in the Restatement list
was:
only the beginning of the enquiry, because although it is easy to say that
some policies of prolonged arbitrary detentions are so bad that those who
enforce them become enemies of the human race, it may be harder to say
which policies cross that line with the certainty afforded by Blackstone’s
three common law offenses.170
This passage implies that even if a CIL norm is included in the Restatement, the norm might not be sufficiently well defined to support a
cause of action under the ATS.
Another way in which the Court limited the sources that had been
relied on in pre-Sosa cases was by shrinking the allowable gap between actual state practice and the proposed CIL cause of action. The
Second Circuit in Filartiga had reasoned that “[t]he fact that the prohibition of torture is often honored in the breach does not diminish its
binding effect as a norm of international law.”171 While acknowledging that “violations of a rule [do not] logically foreclose the existence of
that rule as international law,” the Court in Sosa observed: “that a rule
as stated is as far from full realization as the one Alvarez urges is evidence against its status as binding law; and an even clearer point
against the creation by judges of a private cause of action to enforce
the aspiration behind the rule claimed.”172 Contrary state practice by
itself does not disprove the existence of a rule of CIL or a private
cause of action, but courts will be less likely to recognize the rule or
cause of action when there is a large gap between it and actual state
practice.
In sum, the Court in Sosa departed in many respects from the
lower courts’ prevailing approach to recognizing CIL causes of action
in ATS cases. Its “definition” requirement demands a tight connection
between the plaintiff’s allegations and the sources supporting a CIL
cause of action. And its “acceptance” requirement contemplates a nar–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
169 See, e.g., Hilao v. Estate of Marcos, 103 F.3d 789, 795 (9th Cir. 1996); Presbyterian Church
of Sudan v. Talisman Energy, Inc., 244 F. Supp. 2d 289, 305 (S.D.N.Y. 2003). The Restatement
provides that a state violates CIL if:
as a matter of state policy, it practices, encourages, or condones (a) genocide, (b) slavery
or slave trade, (c) the murder or causing the disappearance of individuals, (d) torture or
other cruel, inhuman, or degrading treatment or punishment, (e) prolonged arbitrary detention, (f) systematic racial discrimination, or (g) a consistent pattern of gross violations
of internationally recognized human rights.
RESTATEMENT (THIRD) OF THE FOREIGN RELATIONS LAW OF THE UNITED STATES § 702
(1987).
170 Sosa, 124 S. Ct. at 2769.
171 Filartiga v. Pena-Irala, 630 F.2d 876, 884 n.15 (2d Cir. 1980).
172 Sosa, 124 S. Ct. at 2769 n.29.
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
901
row conception of relevant sources of law. More specifically, as compared to pre-Sosa practice, the Court in Sosa gave little weight to both
the Universal Declaration of Human Rights and the ICCPR, narrowed
the relevance of national constitutions and the Restatement, and reduced the allowable gap between a CIL norm’s aspiration and the actual practice of states. It is no surprise, in this light, that the Court in
Sosa envisioned that, under its approach, only a modest number of
claims would be recognized under the ATS.173
It remains unclear, however, precisely how far Sosa went in this regard. The lack of clarity results from the Court’s favorable citation to
prior lower court opinions that had embraced the very methods and
sources of CIL identification that the Court in Sosa appeared to discount. For example, the Court stated that its definition and acceptance limitations were “generally consistent” with the reasoning in Filartiga, even though Filartiga relied on sources — General Assembly
resolutions, unratified or non-self-executing treaties, and a survey of
national constitutions — that the Court in Sosa discounted or deemed
irrelevant.174 The Court also cited Filartiga for the proposition that its
position on recognizing CIL causes of action “has been assumed by
some federal courts for 24 years.”175 But Filartiga did not view the
ATS as authorizing the development of federal common law causes of
action. Instead, Filartiga assumed for the purposes of its analysis that
the ATS was a purely jurisdictional statute and was constitutional for
suits between aliens because CIL was federal common law; it did not
consider the cause of action question and indeed the court remanded
on the issue of governing law.176 The Court’s citations to Filartiga are
all the more puzzling because the Court disapprovingly referred to
other cases that had relied on the same sources for the identification of
CIL as those relied on in Filartiga.177 This leaves two possible explanations for the Filartiga references. The Court may have cited Filartiga and related decisions simply for the proposition that some aspects
of CIL could be domesticated in ATS cases, a point on which Filartiga
and Sosa clearly agree. Or it may have cited Filartiga and related decisions not because it agreed with their use of sources, but because it
agreed with their ultimate conclusions about particular rules of CIL.
Regardless of what one makes of these citations, ultimately they must
bear less weight than the Court’s own treatment of controversial
sources of CIL, which, as discussed above, was significantly restrained.
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
173
174
175
176
177
See id. at 2761, 2764.
See id. at 2765–66.
Id. at 2765.
See Filartiga, 630 F.2d at 889.
See Sosa, 124 S. Ct. at 2768 n.27 (disapproving of cases cited in Brief for Respondent Alvarez-Machain at 49 n.50, Sosa, 124 S. Ct. 2739 (No. 03-339)).
902
HARVARD LAW REVIEW
[Vol. 120:869
D. CIL’s Post-Erie Domestic Status in the Absence
of Political Branch Authorization
We now turn to the implications of Sosa for the modern position
that all of CIL, “whatever [it] requires,” is automatically incorporated
wholesale into post-Erie federal common law.178 Recall that the modern position rests on two principal arguments: first, the historical claim
that CIL was part of federal law rather than general law prior to Erie;
and second, the doctrinal claim that CIL was incorporated wholesale
into federal common law after Erie.179 We have already explained
how Sosa rejected the modern position’s historical claim. In this section, we explain why the Court’s reasoning in Sosa cannot be reconciled with the modern position claim about CIL’s post-Erie status.
Our basic argument is that the opinion in Sosa is preoccupied with
the limitations that Erie imposes on the federal courts’ common law–
making powers — limitations that are inconsistent with the modern
position. In particular, the Court’s insistence in Sosa that any federal
common law relating to CIL be grounded in, conform to, and not exceed the contours of what the political branches have authorized; its
recognition that the ATS authorizes courts to enforce only a very small
subset of CIL; and its limited view of judicial power vis-à-vis the federal political branches and even the states in cases involving CIL simply cannot be reconciled with the modern position that all of CIL is
automatically part of judge-made federal common law even in the absence of political branch authorization. After making these points, we
consider counterarguments.
1. Inconsistencies Between Sosa and the Modern Position. — The
Court in Sosa embraced a conventional understanding of the nature of
post-Erie federal common law. It noted that Erie significantly narrowed the common law powers of federal courts to “havens of specialty,” or “interstitial areas of particular federal interest.”180 Although
the Court acknowledged that Sabbatino had recognized a “competence
to make judicial rules of decision of particular importance to foreign
relations, such as the act of state doctrine,” the Court explained that
even in the foreign relations context “the general practice has been to
look for legislative guidance before exercising innovative authority
over substantive law.”181
The Court in Sosa further suggested that these limitations on postErie federal common law had certain implications for judicial incorporation of CIL — three concerning separation of powers and one con–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
178
179
180
181
Brilmayer, supra note 2, at 324 (emphasis omitted).
See supra section III.A, pp. 882–85; section III.B, pp. 885–87.
Sosa, 124 S. Ct. at 2762.
Id.
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
903
cerning federalism. First, and most fundamentally, in deciding
whether the federal courts could incorporate CIL even though the ATS
was merely a jurisdictional statute, all nine Justices engaged in a
search for positive authority for the incorporation. That is, the Court
unanimously turned not to a discussion of powers inherent in the federal courts or of the broad common law status of CIL, but rather to
Congress’s intent in enacting the ATS. Indeed, the entire thrust of the
Sosa opinion was an attempt to ground its holding about the incorporation of CIL in what Congress intended and authorized.182 The historical section of the opinion was consumed by a search for the original understanding of what Congress authorized,183 and the Court built
on this historical understanding to ascertain what the ATS authorized
in modern times. So, for example, the Court rejected the claim that
the ATS, with its “limited, implicit sanction”184 of judicial recognition
of CIL claims, “should be taken as authority to recognize the right of
action asserted by Alvarez here.”185 Rather, it stated as a reason for
judicial caution in creating new causes of action under the ATS that
“[w]e have no congressional mandate to seek out and define new and
debatable violations of the law of nations.”186
Second, and related to the authorization requirement, the Court
made clear that the development of a domestic federal common law of
CIL must proceed interstitially and conform to the wishes of the political branches. The Court twice invoked the non-self-execution declaration attached by the United States to its ratification of the ICCPR
as a reason not to rely on the ICCPR in developing domestic federal
common law related to CIL.187 It also noted that the “affirmative authority” in the Torture Victim Protection Act, which Alvarez-Machain
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
182 Justice Scalia’s concurrence understood the majority’s approach in this way, describing it as
finding “authorization in the understanding of the Congress that enacted the ATS.” Id. at 2773
(Scalia, J., concurring).
183 See id. at 2754 (majority opinion) (“[A]t the time of enactment the [ATS] enabled federal
courts to hear claims in a very limited category defined by the law of nations and recognized at
common law.” (emphasis added)); id. (noting that the ATS gave “limited, implicit sanction to entertain the handful of international law cum common law claims understood in 1789” (emphasis
added)); id. at 2756 (“It was this narrow set of violations of the law of nations, admitting of a judicial remedy and at the same time threatening serious consequences in international affairs, that
was probably on minds of the men who drafted the ATS with its reference to tort.” (emphasis
added)); id. at 2759 (“Congress intended the ATS to furnish jurisdiction for a relatively modest set
of actions alleging violations of the law of nations.” (emphasis added)); id. (“[T]he ATS was meant
to underwrite litigation of a narrow set of common law actions derived from the law of nations . . . .” (emphasis added)); id. at 2761 (“[The ATS] is best read as having been enacted on the
understanding that the common law would provide a cause of action for the modest number of
international law violations with a potential for personal liability at the time.” (emphasis added)).
184 Id. at 2754 (emphasis added).
185 Id. (emphasis added).
186 Id. at 2763.
187 See id. at 2763, 2767.
904
HARVARD LAW REVIEW
[Vol. 120:869
invoked in support of his claim, “is confined to specific subject matter”
and would not support “judicial creativity” beyond its terms.188 And it
tethered modern recognition of CIL-based claims to the historical
paradigms Congress anticipated in enacting the ATS. Finally, the
Court specifically mentioned the possibility of “case-specific deference
to the political branches” in deciding whether to allow particular suits
under the ATS.189
Third, the Court emphasized that the separation of powers limitations on the common law powers of federal courts were especially
forceful in foreign relations cases. As the Court noted, “the potential
implications for the foreign relations of the United States of recognizing [federal common law causes of action in ATS cases] should make
courts particularly wary of impinging on the discretion of the Legislative and Executive Branches in managing foreign affairs.”190
Fourth, the Court explained that Erie’s federalism concerns were
relevant even to a federal common law of CIL. In disputing Justice
Scalia’s contention that the Court’s analysis would mean that the federal question jurisdiction statute, like the ATS, would carry “with it an
opportunity to develop common law,” the Court noted that “our holding today [about the ATS] is consistent with the division of responsibilities between federal and state courts after Erie as a more expansive
common law power related to 28 U.S.C. § 1331 might not be.”191
Even with regard to CIL, the Court made plain, the federalism justification for a narrow reading of the common law powers of federal
courts remains.
This approach to judicial incorporation of CIL is fatal to the modern position that all of CIL is federal common law. As an initial matter, the application of all of CIL as federal common law is inconsistent
with the requirement that federal common law be interstitial. There is
also no political branch authorization for the application of all of CIL
as federal common law. A wholesale incorporation of the CIL of human rights in particular would defy the wishes of the political
branches, which have consistently ratified human rights treaties on the
condition that they not apply as domestic law. The wholesale judicial
incorporation of CIL into domestic federal law is also at odds with the
proposition that federal courts must act cautiously in this area. Finally, to the extent that the federal common law of CIL in the human
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
188 Id. at 2763. The Court drew this conclusion even though the TVPA’s legislative history endorsed ATS litigation, reasoning that “Congress as a body ha[d] done nothing to promote such
suits.” Id. (emphasis added).
189 Id. at 2766 n.21.
190 Id. at 2763; see also id. (“[T]he possible collateral consequences of making international
rules privately actionable argue for judicial caution.”).
191 Id. at 2765 n.19 (citation omitted).
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
905
rights context would address traditional domestic concerns like the
domestic regulation of the death penalty,192 it would be in tension with
Erie’s federalism principles.
Not only is the modern position inconsistent with the post-Erie approach articulated in Sosa, but it is also inconsistent with what the
Court in Sosa stated it was allowing. Sosa involved a situation, according to the Court, in which the political branches had authorized
the incorporation of CIL (through the ATS). Even so, the Court made
clear that it was authorized to incorporate only a small portion of CIL.
In response to Justice Scalia’s claim that Erie precluded the domestic
incorporation of CIL in ATS cases, the Court insisted only that “the
door to further independent judicial recognition of actionable international norms” was not shut altogether,193 but rather was “still ajar subject to vigilant doorkeeping, and thus open to a narrow class of international norms today.”194 The Court repeatedly made the point that it
was not embracing the wholesale incorporation of CIL. Instead, it
was simply defending the possibility that federal courts, consistent
with what Congress has authorized, need not “avert their gaze entirely
from any international norm intended to protect individuals,”195 that
Erie did not cause courts to lose “all capacity to recognize enforceable
international norms,”196 and that nothing Congress had done “shut the
door to the law of nations entirely.”197 These passages assume that
Erie significantly narrowed, but did not eliminate, the circumstances
in which CIL could be applied domestically, consistent with political
branch authorization. The passages are difficult to square with the
modern position claim that all of CIL applies as federal common law
in the absence of political branch authorization.
Finally, the Court’s disclaimer concerning the federal question jurisdiction statute further supports the view that it rejected the modern
position. The Court disagreed with Justice Scalia’s claim that under
the majority’s approach every grant of jurisdiction — most notably the
federal question statute, 28 U.S.C. § 1331 — would carry with it the
opportunity to develop federal common law related to CIL.198 The
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
192 Cf. William A. Schabas, International Law and Abolition of the Death Penalty, 55 WASH. &
LEE L. REV. 797, 799 (1998) (“While it is still premature to declare the death penalty prohibited
by customary international law, it is clear that we are somewhere in the midst of such a process,
indeed considerably close to the goal.”).
193 Sosa, 124 S. Ct. at 2764.
194 Id. (emphasis added). The Court made the same point in noting that its dispute with Justice Scalia’s approach was over “whether some norms of today’s law of nations may ever be recognized legitimately by federal courts in the absence of congressional action beyond § 1350.” Id.
195 Id. at 2764–65 (emphasis added).
196 Id. at 2765 (emphasis added).
197 Id. (emphasis added).
198 Id. at 2765 n.19.
906
HARVARD LAW REVIEW
[Vol. 120:869
Court observed that in contrast to its conclusion that the ATS “was
enacted on the congressional understanding that courts would exercise
jurisdiction by entertaining some common law claims derived from the
law of nations,” it had “no reason to think that federal-question jurisdiction was extended subject to any comparable congressional assumption.”199 The reason there was no comparable congressional assumption was that — as Sosa’s historical analysis suggests, and as we
further explain below200 — CIL was not considered federal law and
thus was not covered by § 1331.
In addition to ruling out the modern position, Sosa also appears to
rule out the “intermediate approach” that has been suggested by some
scholars, whereby U.S. courts would apply CIL as a form of nonfederal or nonpreemptive law, akin to the pre-Erie general common
law.201 As we have already explained, the Court reiterated Erie’s assertion that federal courts could no longer apply general common
law.202 The Court specifically observed that “we now adhere to a conception of limited judicial power first expressed in reorienting federal
diversity jurisdiction [in Erie], that federal courts have no authority to
derive ‘general’ common law.”203 In addition, the Court’s narrow allowance of an incorporation of CIL in discrete circumstances authorized by the political branches differs substantially from the automatic,
unauthorized incorporation of CIL that took place under the general
common law regime.204 Finally, in referring to the causes of action
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
199
200
201
202
Id.
See infra section V.A.1, pp. 911–14.
See supra pp. 886–87.
Justice Scalia’s concurrence similarly described Erie as having categorically disallowed
federal court application of general common law. See Sosa, 124 S. Ct. at 2770–71 (Scalia, J.,
concurring).
203 Id. at 2764 (majority opinion). It would be theoretically consistent with Sosa for federal
courts to apply CIL even in the absence of federal authorization if there were pertinent state law
authorization. Such an authorization would not likely flow, as some have suggested, see, e.g.,
Weisburd, supra note 3, at 52–55, from state choice-of-law rules, which in general look to foreign
law rather than international law. Authorization might conceivably flow from state receiving
statutes, which incorporate into state rules of decision at least part of the common law of England
and might thereby include CIL. See Bradley & Goldsmith, supra note 3, at 870 n.345; Harold H.
Sprout, Theories as to the Applicability of International Law in the Federal Courts of the United
States, 26 AM. J. INT’L L. 280, 282–85 (1932). It is doubtful, however, that such receiving statutes authorize the incorporation of modern CIL, see Bradley & Goldsmith, supra note 3, at 870
n.345, and no state court that we are aware of has construed a receiving statute in that fashion.
The important point for present purposes is that any CIL applied by a federal court pursuant to a
state receiving statute would not be general common law, but rather would be properly characterized as state law. It therefore would not support the “general common law” intermediate position.
204 A subtle constitutional point also confirms this reading of the majority opinion. In citing
favorably to Filartiga and in declining to close the door entirely on suits involving foreign government conduct, the majority appeared to envision that some cases between aliens could properly be maintained under the ATS. If CIL were merely general common law, however, and if (as
the Court unanimously concluded) the ATS is simply a jurisdictional statute, then there would be
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
907
that would be allowed under the ATS, it referred specifically to “federal common law,” not general common law.205
2. Counterarguments. — Commentators who have concluded that
Sosa embraced the modern position tend to ignore the points we have
just discussed and rely instead on two other aspects of the Sosa decision. We consider these aspects below and conclude that they do not
contradict our conclusions.
First, commentators have emphasized the Court’s statement that
“[f]or two centuries we have affirmed that the domestic law of the
United States recognizes the law of nations.”206 As we have explained,
the Court in Sosa understood that CIL did not have the status of federal law before Erie, so this historical statement cannot be a claim
about CIL’s status as federal common law. Indeed, it is telling that the
Court merely stated that U.S. domestic law “recognizes” CIL and did
not claim that CIL is automatically incorporated into U.S. law, let
alone into U.S. federal law. Instead of endorsing the modern position,
the Court was claiming only that U.S. law can take account of CIL,
which is clearly correct independent of the modern position. There
were many instances prior to Erie in which federal courts incorporated
or took account of CIL, and, as we explain in Part V, there are many
post-Erie examples in which federal courts, consistent with the Court’s
statement, borrow from, or “recognize,” CIL in select instances, something far short of the wholesale incorporation of CIL into U.S. federal
law posited by the modern position.
Second, commentators have emphasized two references by the
Court to Sabbatino.207 In a parenthetical, the Court quoted the following dictum from Sabbatino: “[I]t is, of course, true that United States
courts apply international law as a part of our own in appropriate circumstances.”208 The Court also observed in a footnote that Sabbatino
endorsed the reasoning of a short essay published by Professor Philip
Jessup in 1939 that argued that the Erie doctrine should not be ap–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
no basis under Article III of the Constitution for hearing claims between aliens based upon CIL.
See Mossman v. Higginson, 4 U.S. (4 Dall.) 12 (1800) (stating that suits between aliens do not fall
within the Article III Diversity Clause). This Article III problem is addressed, however, if the
claim has the status of federal common law, because federal common law (unlike general common
law) is considered part of the “Laws of the United States” for purposes of Article III. As for the
intermediate position pursuant to which courts would apply CIL as federal law only for purposes
of jurisdiction, see supra pp. 886–87, we would simply note that the Supreme Court even before
Erie did not view CIL as having this status, see supra note 74.
205 Sosa, 124 S. Ct. at 2765.
206 Id. at 2764; see also Dodge, supra note 7, at 95–96; Neuman, supra note 7, at 129;
Steinhardt, supra note 6, at 2252.
207 See, e.g., Neuman, supra note 7, at 129–30.
208 Sosa, 124 S. Ct. at 2764 (alteration in original) (quoting Banco Nacional de Cuba v. Sabbatino, 376 U.S. 398, 423 (1964)) (internal quotation marks omitted).
908
HARVARD LAW REVIEW
[Vol. 120:869
plied to CIL.209 Whatever their precise import, these references are
not an endorsement of a wholesale federalization of CIL. The quotation from Sabbatino refers to the application of CIL “in appropriate
circumstances,” and the Jessup essay merely argues that CIL should
not be treated under Erie as state law, not that it should be treated as
post-Erie federal common law (the conception of which was still in its
infancy in 1939). Moreover, as the Court noted in Sosa, the Sabbatino
decision did not even involve the application of CIL.210 Rather, Sabbatino involved the application of the act of state doctrine, which the
Court made clear was based on domestic separation of powers considerations and was not required by international law.211 While the receiver in Sabbatino raised a CIL argument (concerning a prohibition
on the expropriation of alien property), the Court declined to consider
that argument and indeed declined to apply CIL at all. It is not surprising, therefore, that the Court in Sosa relied on the language in
Sabbatino only for the modest proposition that federal courts need not
“avert their gaze entirely” from CIL norms protecting individuals.212
The Court’s two other references to Sabbatino, which have not
been emphasized by modern position proponents, further confirm that
the Court did not view Sabbatino as support for the modern position.
These references occurred in the course of explaining why courts
should exercise restraint in considering new causes of action based on
CIL.213 One of these references approved Sabbatino’s use of the act of
state doctrine to prevent foreign relations problems that would have
resulted from applying CIL as a rule of decision.214 The Court in Sosa
believed that similar restraint was required in ATS suits that involve
attempts by aliens to enforce international law against their own governments and thus “raise risks of adverse foreign policy consequences.”215 This reliance on Sabbatino as a restraint on the consideration of international claims contradicts the argument that the Court
viewed Sabbatino as supporting the unrestrained incorporation of CIL
into federal common law.
The other reference to Sabbatino occurred in a discussion of Erie’s
transformation of the role of federal courts and of the nature of federal
common law.216 In that context, the Court noted that, “although [it
has] even assumed competence to make judicial rules of decision of
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
209
210
211
212
213
214
215
216
Id. at 2765 n.18 (citing Jessup, supra note 80).
See id.
See Sabbatino, 376 U.S. at 421–23.
Sosa, 124 S. Ct. at 2764–65.
See id. at 2762–63.
See id. at 2762.
Id. at 2763.
See id. at 2762.
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
909
particular importance to foreign relations,” as it did in Sabbatino, “the
general practice has been to look for legislative guidance before exercising innovative authority over substantive law.”217 The Court next
stated, referring to the incorporation of CIL in ATS cases, that it
would “be remarkable to take a more aggressive role in exercising a
jurisdiction that remained largely in shadow for much of the prior two
centuries.”218 This understanding of the import of Sabbatino simply
cannot support the wholesale incorporation of CIL as federal common
law.
A final counterargument against our interpretation of Sosa might
be that the Court’s reasoning is relevant only to whether CIL supports
a domestic cause of action, but has no implications for CIL’s domestic
status in other contexts, such as its ability to preempt state law or
serve as a basis for federal question jurisdiction. This argument is not
open to those who believe that Sosa addressed, and embraced, the
modern position. In any event, we do not believe that it is a fair reading of the Sosa opinion. The Court’s elaborate analysis of Erie and
the nature of post-Erie federal common law would have been largely
unnecessary if the Court had been simply speaking to the circumstances under which courts may imply a cause of action from existing
domestic law, an issue generally governed by different precedent. Indeed, the Court referred to limits on implied rights of action as only
one of many reasons for judicial caution in allowing claims under the
ATS.219 Moreover, for all the reasons discussed earlier, the Court’s
view of post-Erie federal common law is inconsistent with the proposition that CIL automatically and in a wholesale fashion has the status
of federal law, even outside the context of causes of action.
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
217
218
219
Id.
Id.
See id. at 2763.
910
HARVARD LAW REVIEW
[Vol. 120:869
Table 2 illustrates how Sosa resolved the four debates discussed in
Part III:
TABLE 2
Pre-Erie
Status of CIL
Post-Erie
Status of CIL
Nature of ATS
Scope and
Sources of CIL
to be Applied
by Courts in
ATS Litigation
Conventional
Wisdom in 1990s
Federal law
Revisionist View
General law
Sosa
General law
Wholesale
incorporation
as federal
common law
Either creates
federal causes
of action or
authorizes courts
to create them
Selective incorporation
based on constitutional
or political branch
authorization
Only jurisdictional
All of CIL,
derived from wide
range of materials
Limited set
of CIL norms, based
primarily on the
practice of nations
Selective incorporation
based on constitutional
or political branch
authorization
Only jurisdictional,
but nevertheless has
the effect today of
authorizing courts to create some federal causes
of action
Limited set
of CIL norms,
with increased
emphasis on the
practice of nations
V. A POST-SOSA APPROACH TO CIL AS FEDERAL COMMON LAW
We argued in Part IV that Sosa rejected the modern position that
CIL is incorporated wholesale into domestic federal common law. It
does not follow, however, that CIL is irrelevant to federal common law
outside the context of the ATS. The essential problem with the modern position is that it ignores the significance of Erie and the requirements and limitations of post-Erie federal common law. As Sosa made
clear, however, CIL can be incorporated into, or inform, federal common law consistent with the requirements of Erie. In this Part, our
goal is to sketch a general account of the federal common law of CIL
that is more faithful to the premises of post-Erie federal common law
than the overbroad modern position.
We begin by considering possible jurisdictional bases for applying
CIL as federal common law. We next consider a variety of nonjurisdictional contexts in which it may be proper for courts to formulate
federal common law rules relating to CIL, either as gap filling for statutes or treaties, or as the result of certain structural constitutional con-
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
911
siderations. We conclude by considering several areas of likely debate
during the next decade concerning the domestic application of CIL.
A. Possible Jurisdictional Bases for CIL as Federal Common Law
The Supreme Court has made clear that, as a general matter, “[t]he
vesting of jurisdiction in the federal courts does not in and of itself
give rise to authority to formulate federal common law.”220 There are
exceptions to this rule. In Lincoln Mills, the Court (controversially)
interpreted a statute conferring federal jurisdiction over suits involving
violations of labor contracts to authorize federal courts to develop substantive federal common law “fashion[ed] from the policy of our national labor laws.”221 And, in Sosa itself, the Court interpreted congressional expectations relating to a jurisdictional statute (the ATS) as
authorizing courts to develop limited federal common law causes of action related to CIL. Nonetheless, consistent with the general rule, the
Court in Sosa made clear that its interpretation of the ATS did not
“imply that every grant of jurisdiction to a federal court carries with it
an opportunity to develop common law.”222 We now explain why two
other jurisdictional provisions — the federal question statute and the
diversity statute — do not confer federal common law–making power,
and we distinguish them from two jurisdictional clauses in Article III
— for interstate disputes and admiralty disputes — that have been
construed as conferring such authority.
1. Federal Question. — Aside from the ATS, the statute most often
invoked as a basis for federal common law related to CIL is the federal question statute, 28 U.S.C. § 1331, which provides district courts
with “original jurisdiction of all civil actions arising under the Constitution, laws, or treaties of the United States.”223 Prior to Sosa, commentators and litigants had argued that cases based on CIL arose under the “laws of the United States” for the purposes of § 1331. Notice
that the argument here is different than under the ATS. Sosa addressed whether the ATS authorized federal courts to develop federal
common law causes of action under CIL. The argument under the
federal question jurisdiction statute, by contrast, is that CIL is part of
the “laws of the United States” within the meaning of that statute. If
this latter claim is true, then CIL not only gives rise to federal jurisdiction, but is itself also part of federal law with potential implications
under the Constitution’s Take Care and Supremacy Clauses.
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
220
221
Tex. Indus., Inc. v. Radcliff Materials, Inc., 451 U.S. 630, 640–41 (1981).
Textile Workers Union v. Lincoln Mills, 353 U.S. 448, 456 (1957). For criticism of the decision, see, for example, Alexander M. Bickel & Harry H. Wellington, Legislative Purpose and the
Judicial Process: The Lincoln Mills Case, 71 HARV. L. REV. 1 (1957).
222 Sosa, 124 S. Ct. at 2765 n.19.
223 28 U.S.C. § 1331 (2000).
912
HARVARD LAW REVIEW
[Vol. 120:869
Before Sosa, most lower courts that had addressed the question
concluded that CIL was not part of the “laws of the United States” for
purposes of the federal question jurisdiction statute and thus could not
be a basis for federal jurisdiction under that statute.224 The Court in
Sosa itself stated that, in contrast with the ATS, there was “no reason
to think that” Congress intended the federal question jurisdiction statute to authorize courts to apply CIL as federal common law.225 The
Court added that the incorporation of CIL as federal common law under the federal question jurisdiction statute might not be consistent
with “the division of responsibilities between federal and state courts
after Erie.”226 For two reasons, such skepticism about CIL and the
federal question statute is warranted.
First, an analysis similar to the one that the Court in Sosa performed on the ATS, as applied to § 1331 and the “ambient law of the
era” at the time that it was enacted, shows that the framers of § 1331
did not view CIL as part of the “laws of the United States.” Section
1331 was enacted in 1875 without substantial debate.227 It was designed to provide a statutory basis for the exercise of federal question
jurisdiction provided for in Article III.228 But in the nineteenth century, when § 1331 was enacted, Article III’s reference to judicial power
over cases arising under the laws of the United States was not viewed
as including the law of nations.229 This conclusion is consistent with
the proposition, confirmed in Sosa, that CIL in the pre-Erie period
was viewed as general common law, not federal law.
Relatedly, in the same year as the enactment of the 1875 statute,
the Supreme Court held that the phrase “laws of the United States” in
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
224 See, e.g., Princz v. Fed. Republic of Germany, 26 F.3d 1166, 1176 (D.C. Cir. 1994); Xuncax
v. Gramajo, 886 F. Supp. 162, 193–94 (D. Mass. 1995); Handel v. Artukovic, 601 F. Supp. 1421,
1426 (C.D. Cal. 1985). But see Deutsch v. Turner Corp., 324 F.3d 692, 718 (9th Cir. 2003); Forti v.
Suarez-Mason, 672 F. Supp. 1531, 1544 (N.D. Cal. 1987).
225 Sosa, 124 S. Ct. at 2765 n.19.
226 Id.
227 See RICHARD H. FALLON, JR., DANIEL J. MELTZER & DAVID L. SHAPIRO, HART &
WECHSLER’S THE FEDERAL COURTS AND THE FEDERAL SYSTEM 857–58 (5th ed. 2003)
[hereinafter HART & WECHSLER’S FEDERAL COURTS].
228 The most widely quoted (and indeed, the only) contemporary statement about § 1331’s
original meaning came from Senator Matthew Carpenter, who asserted that although “[t]he [Judiciary Act] of 1789 did not confer the whole power which the Constitution conferred . . . [the Act of
March 3, 1875 (later, § 1331)] does. . . . [It] gives precisely the power which the Constitution confers — nothing more, nothing less.” 2 CONG. REC. 4986 (1874). The Supreme Court later held
that § 1331 did not confer all of the jurisdiction provided for in the Article III federal question
provision. See Louisville & Nashville R.R. Co. v. Mottley, 211 U.S. 149 (1908).
229 See, e.g., Am. Ins. Co. v. 356 Bales of Cotton, 26 U.S. (1 Pet.) 511, 545–46 (1828) (holding
that a case involving application of admiralty and maritime law — elements of the law of nations
— “does not . . . arise under the Constitution or laws of the United States” within the meaning of
Article III); see also Bradley & Goldsmith, supra note 3, at 824; Jay, supra note 13, at 1309–11;
Weisburd, supra note 73, at 1214–17. See generally Bradley, supra note 89.
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
913
the statute regulating appellate jurisdiction over state court decisions
did not include the law of nations.230 The Court reasoned that it
lacked appellate jurisdiction to review “general laws of war, as recognized by the law of nations applicable to this case” because they do not
involve “the constitution, laws, treaties, or executive proclamations, of
the United States.”231 Many other decisions in the years between the
1875 statute and Erie reached similar conclusions.232 The same wellunderstood and uncontroversial reason why the law of nations was not
part of the “laws of the United States” for statutory appellate jurisdiction — namely, the law of nations’ status as nonfederal general common law — would have applied to the original federal question jurisdiction statute.
As a result, unlike the ATS, § 1331 was not enacted on the understanding that federal courts would be able to hear CIL-based claims
pursuant to § 1331’s jurisdictional grant. Nor is there any indication
that Congress intended to confer authority to incorporate CIL as federal common law through the general federal question statute. As
with the ATS, probative legislative history surrounding conferral of
general federal question jurisdiction is sparse. General federal question jurisdiction was not conferred with any permanence until 1875
and was then subject to a $500 amount in controversy requirement.233
Nothing in the legislative history of the 1875 conferral of general federal question jurisdiction suggests that Congress even considered CIL.
Indeed, recorded legislative debate on the relevant bills did not focus
on the conferral of general federal question jurisdiction at all.234
Second, as we have already seen, one reason why the Court in Sosa
resisted the idea that the “laws of the United States” in § 1331 included
authority to develop CIL through federal common law–making was
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
230
231
232
See N.Y. Life Ins. Co. v. Hendren, 92 U.S. 286 (1875).
Id. at 286–87.
See, e.g., Ker v. Illinois, 119 U.S. 436, 444 (1886) (holding that the question whether forcible
seizure in a foreign country is grounds to resist trial in state court is “a question of common law,
or of the law of nations” that the Supreme Court has “no right to review”); see also Oliver Am.
Trading Co. v. Mexico, 264 U.S. 440, 442–43 (1924); Huntington v. Attrill, 146 U.S. 657, 683
(1892).
233 Act of Mar. 3, 1875, ch. 137, 18 Stat. 470 (1875); see also RUSSELL R. WHEELER & CYNTHIA HARRISON, CREATING THE FEDERAL JUDICIAL SYSTEM 6 (3d ed. 2005) (noting that
Congress started to grant federal jurisdiction over specific types of cases in 1790 but did not confer “general federal-question jurisdiction until 1875”). Broad federal question jurisdiction was
conferred in 1801 by the outgoing Federalist Party but was repealed the following year. See Act
of Feb. 13, 1801, ch. 4, 2 Stat. 89 (1801), repealed by Act of Mar. 8, 1802, ch. 8, 2 Stat. 132 (1802);
see also FELIX FRANKFURTER & JAMES M. LANDIS, THE BUSINESS OF THE SUPREME
COURT 23–28 (1928).
234 See 3 CONG. REC. 2168, 2240, 2275 (1875); H.J., 43d Cong., 2d Sess. 611, 647–48 (1875);
S.J., 43d Cong., 2d Sess. 371–72 (1875); 2 CONG. REC. 4300–04, 4979–88 (1874); FRANKFURTER
& LANDIS, supra note 233, at 65 & n.34, 66–68.
914
HARVARD LAW REVIEW
[Vol. 120:869
that the assertion of such broad federal common law powers might not
be “consistent with the division of responsibilities between federal and
state courts after Erie.”235 It is one thing for federal courts to recognize a limited set of causes of action in suits brought by a narrow class
of plaintiffs based on a statute that references the law of nations, as
the Court did in Sosa. But when federal courts incorporate CIL
wholesale into domestic law — including those aspects of CIL that increasingly regulate functions formerly regulated by states — they move
from the molecular to the molar, in a manner inconsistent with the
limited common law–making powers of federal courts.
In sum, if one performs the same type of analysis under the federal
question jurisdiction statute that the Court in Sosa performed with respect to the ATS, one reaches the conclusion that the framers of the
federal question statute did not intend to authorize the application of
CIL.
2. Diversity. — The ATS analysis in Sosa also raises a question
about whether CIL can be applied as federal common law in diversity
cases under 28 U.S.C. § 1332. The Court in Sosa tried to recapture in
the post-Erie world the relationship between the ATS and the “ambient law of the era” when the ATS was written.236 The diversity statute
was originally enacted at the same time as the ATS. Moreover, the
framers of the diversity statute clearly contemplated that courts exercising diversity and alienage jurisdiction would apply the law of nations in some cases, at least in the sense in which the law of nations included the law merchant and other aspects of the general law related
to commercial transactions. (This was, after all, what Swift v. Tyson
was all about.) If Sosa’s analysis, translating the pre-Erie general
common law that the ATS’s framers thought would apply in ATS
cases into post-Erie federal common law, applied to the diversity statute as well, then one might argue that the diversity statute should constitute authorization for the application of modern CIL as genuine federal common law in cases that have the required diversity of
citizenship and amount in controversy.
For two reasons, we do not believe that this conclusion follows.
The main reason is Erie itself, which Sosa relied on and affirmed. In
overruling Swift, Erie held that diversity jurisdiction was not a basis
for the application of general common law (even general common law,
like the law merchant, that was part of the law of nations) or for otherwise displacing state common law rules.237 Thousands of post-Erie
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
235
236
237
Sosa v. Alvarez-Machain, 124 S. Ct. 2739, 2765 n.19 (2004).
Id. at 2770.
See supra section II.B, pp. 876–77.
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
915
cases have applied what was formerly general common law as state
law, not federal law.
Second, unlike the diversity statute, the ATS provides jurisdiction
over claims arising from a specific body of law — the law of nations
— that was historically part of the general common law. The diversity
statute, by contrast, provides only party-based jurisdiction and evinces
no intent to facilitate particular legal claims. Rather, diversity jurisdiction was intended to secure a certain type of forum for nonresidents
and aliens, independent of the source of the underlying claim.238 The
diversity statute does not pose the post-Erie translation issue addressed in Sosa since the provision of an unbiased forum to qualifying
parties is unaffected by Erie’s shift in the understanding of substantive
law.239
3. Interstate and Admiralty Disputes. — The Supreme Court has
recognized that it is appropriate to exercise federal common law–
making powers related to CIL in two jurisdictional contexts other than
the ATS. The same day that Erie was decided, the Court drew on
principles of CIL to resolve a boundary dispute between Colorado and
New Mexico and made clear that its rule of decision, drawn from CIL,
had the status of federal common law.240 Similarly, the Court has indicated that when it develops common law in its admiralty jurisdiction, that law has the status of federal common law.241
In our view, the use of CIL in these jurisdictional contexts is easier
to justify under traditional principles of federal common law than the
use of CIL in the diversity and federal question contexts. Consider interstate disputes first. The best argument for the development of federal common law as a rule of decision in these cases is that uniquely
federal interests derived from the Constitution demand a federal
rule.242 These cases are expressly contemplated by Article III, they fall
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
238 See Wythe Holt, The Origins of Alienage Jurisdiction, 14 OKLA. CITY U. L. REV. 547, 554–
64 (1989).
239 To the extent alienage jurisdiction was intended to secure a forum in which treaty obligations (such as those under the Treaty of Peace with Britain) would be enforced, respect for the
Supremacy Clause in state courts and general federal question jurisdiction over claims arising
under treaties now assist in achieving that intent. See id. at 554–64.
240 See Hinderlider v. La Plata River & Cherry Creek Ditch Co., 304 U.S. 92, 106–07, 110
(1938); HART & WECHSLER’S FEDERAL COURTS, supra note 227, at 738–39.
241 See, e.g., Pope & Talbot, Inc. v. Hawn, 346 U.S. 406, 409–10 (1953); see also G. Edward
White, A Customary International Law of Torts, 41 VAL. U. L. REV. (forthcoming 2007) (manuscript at 26–27, on file with the Harvard Law School Library) (arguing that admiralty law was
nonpreemptive federal law pre-Erie); Ernest A. Young, Preemption at Sea, 67 GEO. WASH. L.
REV. 273, 326, 347–48 (1999) (noting, and criticizing, the Supreme Court’s attempt “to ‘translate’
the Framers’ conception of maritime law into . . . [the post-Erie] context” by transforming maritime law from general law into federal common law).
242 See Texas v. New Jersey, 379 U.S. 674, 677 (1965); Henry P. Monaghan, The Supreme Court,
1974 Term—Foreword: Constitutional Common Law, 89 HARV. L. REV. 1, 14 (1975) (“[T]he au-
916
HARVARD LAW REVIEW
[Vol. 120:869
within the Supreme Court’s original jurisdiction,243 they cannot practically be decided by the states or under state law given that the states
themselves are parties,244 they are relatively rare,245 and they involve
the resolution of disputes that are directly analogous to disputes between nations. Moreover, both in the Judiciary Act of 1789 and today,
jurisdiction over interstate disputes has been vested exclusively in the
Supreme Court.246
Even with these limiting factors, the Supreme Court is cautious in
stepping into interstate disputes where the creation of common law
may be required.247 When the Court does craft common law to govern
interstate disputes, it takes into account not only constitutional248 but
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
thority to create federal common law springs of necessity from the structure of the Constitution,
from its basic division of authority between the national government and the states.”).
243 U.S. CONST. art. III, § 2, cl. 2; see also Clark, supra note 13, at 1324–25 (recognizing, but
disputing, the conventional notion that federal common law–making authority in interstate disputes derives from Article III’s jurisdictional grant).
244 See Tex. Indus., Inc. v. Radcliff Materials, Inc., 451 U.S. 630, 641 (1981) (noting that, in interstate disputes, “our federal system does not permit the controversy to be resolved under state
law . . . because the interstate or international nature of the controversy makes it inappropriate
for state law to control”); see also Daniel J. Meltzer, The History and Structure of Article III, 138
U. PA. L. REV. 1569, 1607–08 (1990) (citing Hamilton’s understanding that federal jurisdiction
over suits involving states is grounded in the principle that no man should judge his own case).
245 See HART & WECHSLER’S FEDERAL COURTS, supra note 227, at 279–80.
246 28 U.S.C. § 1251(a) (2000); Judiciary Act of 1789, § 13, 1 Stat. 73, 80.
247 See Illinois v. City of Milwaukee, 406 U.S. 91, 93–94, 108 (1972) (denying Illinois’s motion
to invoke the Court’s original jurisdiction while noting that “[i]t has long been [the Supreme]
Court’s philosophy that ‘[its] original jurisdiction should be invoked sparingly’” and that the exclusive grant of interstate dispute jurisdiction is read as “obligatory only in appropriate cases,”
though stating that the Court’s restrictions on its original jurisdiction stem, at least in part, from a
desire to ease the Court’s docket (quoting Utah v. United States, 394 U.S. 89, 95 (1969))); Connecticut v. Massachusetts, 282 U.S. 660, 669 (1931) (citing Missouri v. Illinois, 200 U.S. 496 (1906),
and New York v. New Jersey, 256 U.S. 296 (1921), in explaining that “[t]he governing rule is that
this Court will not exert its extraordinary power to control the conduct of one State at the suit of
another, unless the threatened invasion of rights is of serious magnitude and established by clear
and convincing evidence”); New York, 256 U.S. at 309 (citing and applying the high standard of
Missouri v. Illinois); Missouri, 200 U.S. at 521 (“Before this court ought to intervene [in interstate
disputes] the case should be of serious magnitude, clearly and fully proved, and the principle to be
applied should be one which the court is prepared deliberately to maintain against all considerations on the other side.”); id. at 517–21; HART & WECHSLER’S FEDERAL COURTS, supra note
227, at 301–03 (noting the Supreme Court’s exercise of discretion to refuse to hear even cases
within the Court’s exclusive jurisdiction); see also Oklahoma v. Texas, 258 U.S. 574, 580 (1922)
(adjudicating conflicting claims by Oklahoma, Texas, and the United States that had led to efforts
to mobilize both states’ militias).
248 See Colorado v. New Mexico, 459 U.S. 176, 183 (1982) (respecting the constitutional sovereignty and equality of states in developing the doctrine of equitable apportionment); Connecticut,
282 U.S. at 670 (noting that in suits regarding the competing water rights of states, “principles of
right and equity shall be applied having regard to the ‘equal level or plane on which all the States
stand, in point of power and right, under our constitutional system’” (quoting Wyoming v. Colorado, 259 U.S. 419, 465, 470 (1922))); Clark, supra note 13, at 1322 (noting that many of the rules
developed in interstate disputes “appear to implement the constitutional equality of the states”);
id. at 1323–25, 1328–31.
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
917
also congressional guidance relevant to the dispute.249 Indeed, when
the Court decides issues arising from interstate compacts approved by
Congress, the Court in effect interprets a congressional act, a task well
within the traditional scope of federal common law–making.250 Moreover, to the extent the Court looks to CIL in creating common law
in these cases, it does not directly incorporate CIL into U.S. domestic
law, but rather draws on the narrow component of CIL that governs,
for example, international boundary or water rights to inform the federal common law that governs resolution of interstate disputes.251 In
this sense, the federal common law developed in interstate cases is
doubly narrow: the occasions in which the Court develops federal
common law are rare, and CIL informs domestic federal law in a limited fashion.
An additional factor distinguishes interstate jurisdiction from the
federal question and diversity contexts. For over 200 years, courts
have not perceived a structural need to apply CIL as federal common
law in diversity and federal question jurisdiction cases. By contrast,
even before Erie, the interstate jurisdiction clause was understood to
authorize federal courts to make federal law in the absence of any legislative guidance, subject to subsequent congressional revision.252
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
249 See City of Milwaukee v. Illinois, 451 U.S. 304, 313–15, 317–23 (1981) (noting that Congress
may by subsequent legislation displace federal common law regarding interstate disputes and
holding that Congress had done just that in enacting the comprehensive Federal Water Pollution
Control Act Amendments of 1972); id. at 316–17 (explaining that the standard for finding congressional preemption of federal common law is lower than for finding congressional preemption
of state law); Illinois, 406 U.S. at 101–04 & n.5 (finding that Congress had neither prescribed nor
prohibited the remedy Illinois sought but that the statutes Congress had enacted, while “not necessarily [defining] the outer bounds of the federal common law,” might “provide useful guidelines
in fashioning such rules of decision”); Missouri, 200 U.S. at 518–19 (citing Pennsylvania v. Wheeling & Belmont Bridge Co., 54 U.S. (13 How.) 518 (1851), in which the Court relied on related but
not controlling congressional acts and a congressionally approved interstate compact to resolve an
interstate nuisance dispute); HART & WECHSLER’S FEDERAL COURTS, supra note 227, at 740
(noting the Court’s use of congressional guidance in interstate water pollution disputes, one of few
areas of interstate dispute in which congressional guidance is available).
250 See Wheeling & Belmont Bridge Co., 54 U.S. (13 How.) at 565–66 (noting that an interstate
compact approved by Congress becomes a law enforceable by the Supreme Court); HART &
WECHSLER’S FEDERAL COURTS, supra note 227, at 739.
251 See Meltzer, supra note 146, at 540; see also New Jersey v. Delaware, 291 U.S. 361, 378–85
(1934) (applying the international law doctrine of the Thalweg to resolve a boundary dispute between New Jersey and Delaware); Connecticut, 282 U.S. at 670 (“For the decision of suits between
States, federal, state and international law are considered and applied by this Court as the exigencies of the particular case may require.”); HART & WECHSLER’S FEDERAL COURTS, supra note
227, at 287 (“The Court draws on federal, state and international law, as appropriate, in fashioning the[] common law rules [that govern interstate disputes].”); Clark, supra note 13, at 1328–30
(noting that in implementing the constitutional equality of the states through its interstate dispute
common law, the Court in some cases may “borrow international law doctrines [that
were] originally developed to implement the ‘absolute equality’ of sovereign nations” despite the
fact that these doctrines do not apply to interstate disputes “of their own force”).
252 See Connecticut, 282 U.S. at 670–71; Kansas v. Colorado, 206 U.S. 46, 97 (1907).
918
HARVARD LAW REVIEW
[Vol. 120:869
Claims of structural necessity as a basis for federal common law are
more plausible if these claims have a long historical pedigree.
The same basic analysis applies to admiralty jurisdiction. Admiralty is, of course, a traditional component of the law of nations that
was important to the prosperity of the infant United States.253 Even
the weak national government during the period of Confederation exercised some authority over admiralty disputes.254 When it came time
to craft the new Union, opposition to a broad federal judiciary was
strong.255 The proposed grant of federal diversity jurisdiction, for example, was bitterly opposed.256 By contrast, even those who opposed
the federal judicial system contemplated by the Constitution and established by the Judiciary Act agreed on the need for “national admiralty courts.”257 “When proposals to abolish Congress’s Article III authority to establish federal courts were made in the state ratifying
conventions and in the First Congress, there was usually an exception
for courts of admiralty.”258 As a result, and in notable contrast with
the treatment of the law of nations more generally, the Constitution
explicitly extended federal judicial authority to include admiralty.259
Further, Congress has enacted various statutes to govern private
admiralty issues. Thus, much of federal admiralty law today is found
in statutes or treaties and not exclusively in the common law.260 CIL
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
253 See HART & WECHSLER’S FEDERAL COURTS, supra note 227, at 15; David P. Currie,
Federalism and the Admiralty: “The Devil’s Own Mess,” 1960 SUP. CT. REV. 158, 163–64; John P.
Frank, Historical Bases of the Federal Judicial System, 13 LAW & CONTEMP. PROBS. 3, 14
(1948).
254 See HART & WECHSLER’S FEDERAL COURTS, supra note 227, at 6 & nn.32–33 (describing the Continental Congress’s authority “to ‘appoint’ state courts for the trial of ‘piracies and
felonies on the high seas’” and Congress’s establishment of a national tribunal to hear appeals in
capture cases); Frank, supra note 253, at 6–9 (describing admiralty courts during the colonial and
Confederation periods).
255 See HART & WECHSLER’S FEDERAL COURTS, supra note 227, at 6–9.
256 See id. at 17, 19; Wythe Holt, “To Establish Justice”: Politics, the Judiciary Act of 1789, and
the Invention of the Federal Courts, 1989 DUKE L.J. 1421, 1468–71, 1477; Charles Warren, New
Light on the History of the Federal Judiciary Act of 1789, 37 HARV. L. REV. 49, 56 (1923).
257 WHEELER & HARRISON, supra note 233, at 6; see also Michael G. Collins, The Federal
Courts, the First Congress, and the Non-Settlement of 1789, 91 VA. L. REV. 1515, 1520–21, 1523–
30, 1539–40, 1555, 1565, 1570–71 (2005); Frank, supra note 253, at 9; Henry J. Friendly, The Historic Basis of Diversity Jurisdiction, 41 HARV. L. REV. 483, 484 n.6 (1928); Holt, supra note 256,
at 1428–30 & n.26; Young, supra note 241, at 277–80 & nn.41–42, 314–17, 348 (noting a consensus
in favor of federal admiralty jurisdiction, though disputing the suggestion that the Framers intended to federalize all substantive admiralty law).
258 WHEELER & HARRISON, supra note 233, at 6; see also Warren, supra note 256, at 119–20
(describing such a proposal in the First Congress); cf. id. at 123 & n.166 (noting a similar attempt
to amend what became the First Judiciary Act to limit lower federal courts to acting as courts of
admiralty).
259 See U.S. CONST. art. III, § 2, cl. 1.
260 See Miles v. Apex Marine Corp., 498 U.S. 19, 27 (1990); HART & WECHSLER’S FEDERAL
COURTS, supra note 227, at 735; Robert Force, An Essay on Federal Common Law and Admiralty, 43 ST. LOUIS U. L.J. 1367, 1370–77, 1382–84 (1999); Jonathan M. Gutoff, Federal Common
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
919
is often used for interstitial gap-filling, an uncontroversial use of federal common law261 that is a far cry from the wholesale incorporation
of CIL contemplated by those who advocate the use of the federal
question statute or the diversity statute as a basis for treating CIL as
federal common law. Even on issues with respect to which Congress
has not specifically legislated, the Supreme Court has tried to conform
the common law of admiralty to Congress’s intent behind related statutes.262 In short, admiralty is only one small subset of CIL, and it is a
subset in which federal common law is used selectively to promote the
policies adopted by the political branches.
B. Possible Substantive Bases for CIL as Federal Common Law
We now turn from an examination of possible jurisdictional authorizations to possible substantive authorizations — in statutes, treaties, and executive branch pronouncements — for a federal common
law of CIL. As we explain, there continues to be a robust role for CIL
in the U.S. legal system even if one rejects the modern position.
1. Statutes. — Some statutes specifically reference CIL and thus
invite courts to incorporate and interpret CIL as part of the statutory
scheme. An oft-cited example is the federal piracy statute, which pro–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Law and Congressional Delegation: A Reconceptualization of Admiralty, 61 U. PITT. L. REV. 367,
374 & n.32, 405–06 (2000); Ernest A. Young, It’s Just Water: Toward the Normalization of Admiralty, 35 J. MAR. L. & COM. 469, 477 & n.31 (2004); cf. Gutoff, supra, at 403–06, 417 (finding congressional delegation of authority to create a federal common law of admiralty in Congress’s reenactment and expansion of admiralty jurisdiction in 1948).
261 See Young, supra note 241, at 477.
262 See Norfolk Shipbuilding & Drydock Corp. v. Garris, 532 U.S. 811, 820 (2001) (citing
American Dredging Co. v. Miller, 510 U.S. 443 (1994), for the proposition that federal common
law–making in admiralty should “harmonize with the enactments of Congress in the field”); id. at
821 (Ginsburg, J., concurring in part) (same); Am. Dredging Co., 510 U.S. at 456–57 (following the
Jones Act’s lead in finding that state forum non conveniens rules may apply to general maritime
claims); Miles, 498 U.S. at 27 (“In this era [in which Congress has legislated extensively on admiralty matters], an admiralty court should look primarily to these legislative enactments for policy
guidance.”); id. at 32–37 (limiting recovery under general maritime law to coincide with the limited recovery sanctioned by Congress in related statutes); Moragne v. States Marine Lines, Inc.,
398 U.S. 375, 390–402 (1970) (noting that legislative “policy carries significance beyond the particular scope of . . . the statutes involved” and should “be given its appropriate weight . . . in matters of . . . decisional law” and relying on the policies behind related, but not controlling, federal
statutes to recognize a wrongful death remedy in general maritime law); S. Pac. Co. v. Jensen, 244
U.S. 205, 218 (1917) (bolstering the conclusion that New York’s Workmen’s Compensation Act
did not apply to a maritime accident by noting that the workmen’s compensation remedy would
be inconsistent “with the policy of Congress to encourage investments in ships” as manifested
in two acts “which declare a limitation upon the liability of [ship] owners”); cf. Norfolk Shipbuilding, 532 U.S. at 820 (“[Given] Congress’s extensive involvement in legislating causes of action for
maritime personal injuries, it will be the better course, in many cases that assert new claims
beyond what those statutes have seen fit to allow, to leave further development to Congress.”);
Nw. Airlines, Inc. v. Transp. Workers Union, 451 U.S. 77, 97 n.40 (1981) (“[E]ven in admiralty we
decline to fashion new remedies if there is a possibility that they may interfere with a legislative
program.”).
920
HARVARD LAW REVIEW
[Vol. 120:869
vides that “[w]hoever, on the high seas, commits the crime of piracy as
defined by the law of nations, and is afterwards brought into or found
in the United States, shall be imprisoned for life.”263 This statute
clearly authorizes courts to ascertain and apply as federal law the CIL
prohibition on piracy.264 In this situation, it makes sense to talk about
a federal law status for CIL. Similarly, the Foreign Sovereign Immunities Act265 (FSIA) provides an exception to foreign governmental
immunity for certain situations in which “rights in property taken in
violation of international law are in issue.”266 The phrase “international law” in this exception refers primarily to the CIL governing the
expropriation of alien property. When courts apply a CIL standard
under this jurisdiction, they are best understood as doing so under a
federal common law rationale.267
Sometimes courts develop federal common law not pursuant to an
express reference in a statute, but rather in order to fill gaps in a statutory scheme. For example, some courts have looked to CIL in interpreting aspects of the FSIA, even where CIL is not expressly incorporated, based on indications in the FSIA’s legislative history that this is
what Congress intended.268 Another example comes from the Supreme
Court’s decision in First National City Bank v. Banco Para El Comercio Exterior de Cuba.269 In that case, the issue was what body of law
should apply in determining whether to pierce the veil between a foreign government and its state-owned corporation for purposes of attributing the government’s actions to the corporation (and thereby allowing a counterclaim of expropriation to be brought against the
corporation). The FSIA, which provided the basis for subject matter
jurisdiction and the potential abrogation of sovereign immunity, did
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
263
264
18 U.S.C. § 1651 (2000).
See, e.g., United States v. Smith, 18 U.S. (5 Wheat.) 153 (1820); United States v. Palmer, 16
U.S. (3 Wheat.) 610 (1818).
265 Pub. L. No. 94-583, 90 Stat. 2891 (1976) (codified in scattered sections of 28 U.S.C.).
266 28 U.S.C. § 1605(a)(3) (2000).
267 See, e.g., West v. Multibanco Comermex, S.A., 807 F.2d 820, 831 n.10 (9th Cir. 1987) (“It is
appropriate to look to international law when determining whether [an action] constitutes a ‘taking’ for purposes of FSIA.”); cf. 28 U.S.C. § 1350 note (2000) (defining “extrajudicial killing” in the
Torture Victim Protection Act as not including a killing “that, under international law, is lawfully
carried out under the authority of a foreign nation” (emphasis added)); Department of the Interior,
Environment, and Related Agencies Appropriations Act, Pub. L. No. 109-54, § 201, 119 Stat. 499,
531 (2006) (requiring the EPA to regulate the use of human subjects in pesticide testing “consistent
with . . . the principles of the Nuremberg Code with respect to human experimentation”).
268 See, e.g., Aquamar S.A. v. Del Monte Fresh Produce N.A., Inc., 179 F.3d 1279, 1294–96
(11th Cir. 1999); cf. In re Estate of Ferdinand E. Marcos Human Rights Litig., 978 F.2d 493, 497–
98 (9th Cir. 1992) (“Congress intended the FSIA to be consistent with international law . . . .”);
Tex. Trading & Milling Corp. v. Fed. Republic of Nig., 647 F.2d 300, 310 (2d Cir. 1981) (“[The
FSIA’s] drafters seem to have intended rather generally to bring American sovereign immunity
practice into line with that of other nations.”).
269 462 U.S. 611 (1983).
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
921
not address this issue. In resolving the question, the Supreme Court
developed federal common law based on what it described as “principles . . . common to both international law and federal common law,
which in these circumstances is necessarily informed both by international law principles and by articulated congressional policies.”270 As
we noted earlier, this sort of statutory gap-filling, guided by congressional intent, is probably the most common (and uncontroversial) type
of federal common law.
Applying the Charming Betsy canon of construction is another way
in which courts may look to CIL in the context of statutory construction. Pursuant to this canon, courts construe ambiguous federal statutes to avoid conflicts with international law.271 CIL is not applied as
a rule of decision under this canon, but rather as a relevant consideration in discerning Congress’s intent. This canon almost certainly has
the status of federal common law because a state court interpreting a
federal statute would be bound to follow it. Indeed, the obligation of
state courts to construe a federal statute in the same way that the Supreme Court would construe it (including by reference to the Charming
Betsy canon where relevant) can be seen as the flip side of Erie.272
2. Treaties. — When U.S. courts apply treaties, they sometimes
look to CIL principles to resolve ambiguities and fill in gaps. In doing
so, they often rely on the Vienna Convention on the Law of Treaties,273
which sets forth a variety of general rules governing the formation, interpretation, and termination of treaties. The United States has not
ratified the Convention and thus it cannot bind the United States as a
treaty, but the U.S. government has stated that many of the Convention’s terms reflect CIL.274 Perhaps not surprisingly, therefore, courts
sometimes invoke the CIL of treaty law as embodied in the Vienna
Convention.275 Most often, they apply the principles of interpretation
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
270
271
Id. at 623.
See Murray v. The Schooner Charming Betsy, 6 U.S. (2 Cranch) 64, 118 (1804); see also F.
Hoffman-La Roche Ltd. v. Empagran S.A., 124 S. Ct. 2359, 2366 (2004); Trans World Airlines,
Inc. v. Franklin Mint Corp., 466 U.S. 243, 252 (1984); Weinberger v. Rossi, 456 U.S. 25, 32 (1982);
Lauritzen v. Larsen, 345 U.S. 571, 578 (1953); RESTATEMENT (THIRD) OF THE FOREIGN RELATIONS LAW OF THE UNITED STATES § 114 (1987) (“Where fairly possible, a United States
statute is to be construed so as not to conflict with international law or with an international
agreement of the United States.”).
272 See Curtis A. Bradley, The Charming Betsy Canon and Separation of Powers: Rethinking
the Interpretive Role of International Law, 86 GEO. L.J. 479, 534 n.305 (1997).
273 Opened for signature May 23, 1969, 1155 U.N.T.S. 331.
274 See RESTATEMENT (THIRD) OF THE FOREIGN RELATIONS LAW OF THE UNITED
STATES § 102 n.2 (documenting executive branch statements); S. EXEC. DOC. NO. L, at 1 (1971)
(noting that the Vienna Convention “is already generally recognized as the authoritative guide to
current treaty law and practice”).
275 The Second Circuit recently explained why (and how) it believed it could apply CIL based
on the Vienna Convention even though the United States had not ratified the Convention:
922
HARVARD LAW REVIEW
[Vol. 120:869
in Articles 31 and 32 of the Convention to construe treaties that the
United States has ratified.276 Sometimes, courts look to principles of
CIL as embodied in the Vienna Convention to ascertain whether a
treaty exists.277 It is unclear whether the authorization for courts to
apply the CIL of treaty law in these contexts is best thought of as coming from the ratified treaty in question or from the executive branch.
But in any event, as in the statutory authorization cases, these gapfilling and interpretive uses of CIL are similar to the federal common
law that has been applied in the domestic realm, and these uses are
closely tied to the actions and policies of the political branches.
3. Executive Branch Authorization. — In some circumstances, the
executive branch can provide the authorization for courts to draw
upon CIL in developing federal common law. A particularly good
example is the way in which courts have addressed head-of-state
immunity. For most of our nation’s history, head-of-state immunity
was viewed as a component of foreign sovereign immunity. Prior to
Erie, and consistent with the view that CIL was treated as nonfederal
general common law, federal and state courts alike applied the CIL of
foreign sovereign immunity on the domestic plane without authorization from Congress or the Executive.278 Around the time of Erie, the
Supreme Court stopped applying the CIL of immunity on its own authority, as it had done under the general common law regime, and
began to justify its application on the basis of executive branch authorization.279 The Supreme Court never expressly tied its shift in
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Although we have previously recognized the Vienna Convention as a source of customary international law, it bears underscoring that the United States has never ratified the
Convention. Accordingly, the Vienna Convention is not a primary source of customary
international law, but rather one of the secondary sources “summarizing international
law,” that we rely upon “only insofar as they rest on factual and accurate descriptions of
the past practices of states.”
Avero Belg. Ins. v. Am. Airlines, Inc., 423 F.3d 73, 79 n.8 (2d Cir. 2005) (citations omitted) (quoting Flores v. S. Peru Copper Corp., 414 F.3d 233, 251 (2d Cir. 2003); United States v. Yousef, 327
F.3d 56, 99 (2d Cir. 2003)).
276 See, e.g., Gonzalez v. Gutierrez, 311 F.3d 942, 949 & n.15 (9th Cir. 2002). See generally
Evan Criddle, The Vienna Convention on the Law of Treaties in U.S. Treaty Interpretation, 44 VA.
J. INT’L L. 431 (2004).
277 See, e.g., Chubb & Son, Inc. v. Asiana Airlines, 214 F.3d 301, 307–08 (2d Cir. 2000).
278 Thus, for example, in the 1812 decision The Schooner Exchange v. M’Faddon, 11 U.S. (7
Cranch) 116 (1812), the Supreme Court applied the CIL of sovereign immunity without bothering
to consider domestic authorization to do so. Similarly, in Hatch v. Baez, 14 N.Y. Sup. Ct. 596
(Gen. Term 1876), a New York court relied on an English precedent but no domestic authorization in holding that the former President of the Dominican Republic was entitled to immunity for
his official acts. Id. at 599–600. See generally Julian G. Ku, Customary International Law in
State Courts, 42 VA. J. INT’L L. 265, 307–22 (2001).
279 In Compania Espanola de Navegacion Maritima S.A. v. The Navemar, 303 U.S. 68 (1938),
decided just three months before Erie and issued the day Erie was argued, the Court intimated
for the first time that courts were bound by executive suggestions of immunity. Id. at 74. Subsequently, in its 1943 decision Ex parte Republic of Peru, 318 U.S. 578 (1943), the Court squarely
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
923
treatment of foreign sovereign immunity doctrines to Erie. But the
shift took place at approximately the same time as Erie, and it is
easy to understand why Erie was pivotal: Erie’s positivism required
all applications of law to be grounded in a constitutional or political
branch authorization, and there was no other plausible source of
authorization.280
In 1976, the FSIA transferred the political branch authorization for
judicial application of foreign sovereign immunity from executive suggestion to congressional statute. The FSIA did not specify whether its
immunities extend to heads of state, either current or former.281 This
silence raised the question of whether a foreign head of state was entitled to immunity in U.S. courts after passage of the FSIA, and if so, on
what basis. Although courts have varied in their answers to this question, they have almost always grounded the application of head-ofstate immunity in an authorization from the political branches.282
Some courts view the FSIA as providing for head-of-state immunity,
even though the text of the statute is silent on the issue.283 Most
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
held that, because immunity determinations implicated important foreign relations interests,
courts were bound to follow executive suggestions of immunity. Id. at 587. Two years later, in
Republic of Mexico v. Hoffman, 324 U.S. 30 (1945), the Court went further, stating that even in
the face of executive branch silence, U.S. courts should look to “the principles accepted by the
[executive branch].” Id. at 35. As a result, the Court explained that “it is . . . not for the courts to
deny an immunity which our government has seen fit to allow, or to allow an immunity on new
grounds which the government has not seen fit to recognize.” Id.; see also Curtis A. Bradley &
Jack L. Goldsmith, Pinochet and International Human Rights Litigation, 97 MICH. L. REV. 2129
(1999).
280 This posture was especially appropriate because, at the time of Erie, the CIL of immunity
was in the midst of a transformation that rendered it less amenable to independent judicial determination. During the nineteenth century, the United States, like many other countries, adhered
to the “absolute” theory of sovereign immunity, under which foreign governments were entitled to
immunity for essentially all of their acts, even those that were purely commercial in nature. See
BARRY E. CARTER, PHILLIP R. TRIMBLE & CURTIS A. BRADLEY, INTERNATIONAL LAW
547–50 (4th ed. 2003). In the early twentieth century, however, a number of countries began embracing the “restrictive” theory, under which foreign governments were entitled to immunity for
their public or sovereign acts, but not for their private or commercial acts. See id. at 550–52.
This shift to the restrictive theory, formally endorsed by the U.S. State Department in 1952, see
Letter from Jack B. Tate, Acting Legal Adviser, U.S. Dep’t of State, to Attorney Gen. (May 19,
1952), reprinted in 26 DEP’T STATE BULL. 984 (1952), made the CIL of immunity much more
complex and difficult to apply. It also meant that foreign sovereigns would be haled into court
more often, thereby heightening the foreign policy stakes associated with immunity determinations. In this environment, it made sense that unelected judges with no foreign relations expertise
would seek political branch guidance on whether and how to apply foreign sovereign immunity.
281 The FSIA defines “foreign state” to include a “political subdivision” or an “agency or instrumentality” of a foreign state, 28 U.S.C. § 1603(a) (2000), but neither the statute nor its legislative history mentions head-of-state immunity.
282 Bradley & Goldsmith, supra note 279, at 2166.
283 See, e.g., O’Hair v. Wojtyla, No. 79-2463 (D.C. Cir. Oct. 3, 1979), excerpted in DIGEST OF
UNITED STATES PRACTICE IN INTERNATIONAL LAW 1979, at 897 (Marian Lloyd Nash ed.,
1983).
924
HARVARD LAW REVIEW
[Vol. 120:869
courts, however, view the FSIA as inapplicable to a head of state and
instead look to executive branch authorization to apply the doctrine.284
Among the courts that seek executive branch authorization, some recognize head-of-state immunity only in the face of an explicit suggestion
of immunity by the Executive.285 Others rely on the lack of an executive branch suggestion simply as a factor weighing against immunity.286 In all these cases, the courts are looking, at least to some degree, for political branch authorization.
C. CIL as Federal Common Law: Future Debates
In this section, we examine three contexts in which CIL’s status as
domestic law is likely to be most debated during the next decade. The
first involves corporate liability for alleged human rights violations,
the second involves the war on terrorism, and the third involves the
Supreme Court’s use of foreign and international materials to inform
constitutional interpretation.
1. Corporate Aiding and Abetting Liability. — Some courts prior to
Sosa suggested that corporations could be held liable under the ATS
for aiding and abetting human rights abuses committed by foreign
governments.287 If this proposition were legally correct, it would substantially increase the number and scope of potential ATS cases as
compared with the first wave of ATS cases brought against state officials. Among other things, the number of ATS defendants subject to
personal jurisdiction in the United States would expand; corporations
typically have more assets than individual defendants and thus are
likely to be a more attractive target for plaintiffs and their lawyers;
and private corporations, unlike foreign governments, are not protected by sovereign immunity. For a variety of reasons, we believe
that the best reading of Sosa is that ATS liability cannot be extended
to corporations based on an aiding and abetting theory absent further
action by Congress.
Most international law — both treaty-based and customary — imposes obligations only on States.288 This is true even of much of hu–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
284 See, e.g., Ye v. Zemin, 383 F.3d 620, 624–25 (7th Cir. 2004); United States v. Noriega, 117
F.3d 1206, 1212 (11th Cir. 1997); Doe v. Roman Catholic Diocese, 408 F. Supp. 2d 272, 277–78
(S.D. Tex. 2005); Lafontant v. Aristide, 844 F. Supp. 128, 137 (E.D.N.Y. 1994).
285 See, e.g., Jungquist v. Nahyan, 940 F. Supp. 312, 321 (D.D.C. 1996), rev’d on other grounds,
115 F.3d 1020 (D.C. Cir. 1997).
286 See, e.g., First Am. Corp. v. Al-Nahyan, 948 F. Supp. 1107, 1120–21 (D.D.C. 1996).
287 See, e.g., Doe I v. Unocal Corp., 395 F.3d 932 (9th Cir. 2002).
288 RESTATEMENT (THIRD) OF THE FOREIGN RELATIONS LAW OF THE UNITED STATES
ch. 2, introductory note (1987); 1 OPPENHEIM’S INTERNATIONAL LAW § 6 (Robert Jennings &
Arthur Watts eds., 9th ed. 1992); Carlos M. Vazquez, Direct vs. Indirect Obligations of Corporations Under International Law, 43 COLUM. J. TRANSNAT’L L. 927, 932 (2005).
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
925
man rights law. The Convention Against Torture,289 for example, addresses only torture “by or at the instigation of or with the consent or
acquiescence of a public official or other person acting in an official
capacity.”290 There are a few norms of international law, such as prohibitions on genocide and war crimes, that apply to individuals, at
least for the purpose of criminal prosecution.291 If such norms were
also applicable to corporations (a questionable proposition),292 a corporation could be subject to liability under the ATS for directly violating
one of these norms, assuming the other requirements in Sosa are satisfied.293 Even if a direct liability claim were appropriate, Sosa suggests
that it may be necessary for courts to apply limiting doctrines designed
to promote international comity, such as the act of state doctrine and a
requirement of exhaustion of local remedies.294
Corporations, however, do not typically commit, or even conspire to
commit, genocide or war crimes. As a result, most of the ATS claims
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
289 Convention Against Torture and Other Cruel, Inhuman or Degrading Treatment or Punishment, opened for signature Dec. 10, 1984, 108 Stat. 382, 1465 U.N.T.S. 85.
290 Id. art. 1(1).
291 See Rome Statute of the International Criminal Court art. 25(1), July 17, 1998, 2187
U.N.T.S. 90; Statute of the International Criminal Tribunal for the Former Yugoslavia, S.C. Res.
827, arts. 2–5, U.N. Doc. S/RES/827 (May 25, 1993); Statute of the International Criminal Tribunal for Rwanda, S.C. Res. 955, arts. 2–4, U.N. Doc. S/RES/955 (Nov. 8, 1994). See generally
STEVEN R. RATNER & JASON S. ABRAMS, ACCOUNTABILITY FOR HUMAN RIGHTS
ATROCITIES IN INTERNATIONAL LAW (2d ed. 2001).
292 Cf. Vazquez, supra note 288, at 943–44. It is noteworthy that none of the modern international criminal tribunals extends criminal liability to corporations and that the state parties to the
relatively recent International Criminal Court negotiations considered and rejected international
criminal liability for corporations. See THE INTERNATIONAL CRIMINAL COURT: THE MAKING OF THE ROME STATUTE 198–200 (Roy S. Lee ed., 1999); 1 THE ROME STATUTE OF THE
INTERNATIONAL CRIMINAL COURT 778–79 (Antonio Cassese et al. eds., 2002). Many scholars
nonetheless believe that corporations can be liable under international criminal law. See, e.g.,
Surya Deva, Human Rights Violations by Multinational Corporations and International Law:
Where from Here?, 19 CONN. J. INT’L L. 1 (2003); Louis Henkin, The Universal Declaration at
50 and the Challenge of Global Markets, 25 BROOK. J. INT’L L. 17 (1999); Beth Stephens, The
Amorality of Profit: Transnational Corporations and Human Rights, 20 BERKELEY J. INT’L L.
45 (2002).
293 This direct liability might even extend to some situations involving conspiracy, joint venture, or vicarious liability. See, e.g., Sarei v. Rio Tinto, PLC, 456 F.3d 1069 (9th Cir. 2006) (allowing suit against a corporation to proceed on the theory that it was vicariously liable for human
rights abuses allegedly committed by a foreign government on its behalf).
294 See Sosa v. Alvarez-Machain, 124 S. Ct. 2739, 2766 n.21 (2004); see also id. at 2782–83
(Breyer, J., concurring) (emphasizing the importance of comity considerations in ATS cases). We
thus disagree with the Ninth Circuit’s 2–1 decision in Sarei, in which the court declined to apply
an exhaustion requirement to corporate ATS suits, even though it acknowledged that there was
international law support for such a requirement. That decision also appears to be inconsistent
with Hamdan v. Rumsfeld, 126 S. Ct. 2749 (2006). In Hamdan, the Court held that a statute that
allowed for trial of offenses under international law also implicitly incorporated international law
limitations on such trials, including procedural limitations. See id. at 2794. The ATS’s authorization of civil claims for certain international law violations should also be read as incorporating
international law limitations on such claims.
926
HARVARD LAW REVIEW
[Vol. 120:869
brought against corporations have alleged that they were indirectly liable for human rights abuses committed by foreign government actors
as a result of their acts of aiding and abetting, such as providing the
perpetrators with financial support or materials. There is already a
division in the courts over whether such a common law claim is consistent with Sosa.295 We agree with the courts that have found that it
is not.
As an initial matter, it is important to recall that the text of the
ATS refers to torts “committed” in violation of international law.
There is no suggestion in this language of third-party liability for those
who facilitate the commission of such torts. By contrast, just a year
after the enactment of the ATS, Congress enacted a criminal statute
containing specific provisions for indirect liability — for example, for
aiding or assisting piracy.296
The analysis in Sosa suggests a number of reasons why aiding and
abetting liability should not be read into the ATS. The Court repeatedly emphasized that, consistent with the limited nature of the ATS
and the separation of powers constraints on the federal courts, only a
“modest number” of claims could be brought under the ATS without
further congressional authorization.297 The Court further counseled
the lower courts to exercise “great caution” in recognizing new
claims.298 And the Court emphasized that “innovative” interpretations
should be left to Congress.299 As we noted earlier, however, allowing
corporate aiding and abetting liability would significantly expand ATS
litigation. It would also require courts to exercise significant policy
judgment normally reserved to the legislature, such as fashioning the
precise standards for what constitutes aiding and abetting.
For similar reasons, the Supreme Court declined to imply aiding
and abetting liability in civil cases brought under the securities fraud
statute. In Central Bank of Denver v. First Interstate Bank of Denver,300 the Court reasoned that allowing aiding and abetting liability
for securities fraud would expand litigation in a way that would implicate policy tradeoffs best resolved by Congress.301 The Court also reasoned that Congress’s authorization of aiding and abetting liability
in the criminal context did not suggest a general acceptance of that
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
295 Compare In re S. African Apartheid Litig., 346 F. Supp. 2d 538, 550 (S.D.N.Y. 2004) (disallowing aiding and abetting liability), and Doe I v. Exxon Mobil Corp., 393 F. Supp. 2d 20, 24
(D.D.C. 2005) (same), with Presbyterian Church of Sudan v. Talisman Energy, Inc., 374 F. Supp.
2d 331, 337–38 (S.D.N.Y. 2005) (allowing aiding and abetting liability).
296 See Act of Apr. 30, 1790, ch. 9, § 10, 1 Stat. 112, 114.
297 Sosa, 124 S. Ct. at 2761.
298 Id. at 2763.
299 Id. at 2762.
300 511 U.S. 164 (1994).
301 See id. at 188–89.
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
927
type of liability in the civil context.302 Finally, the Court noted the
substantial uncertainties associated with the standard for aiding and
abetting.303
Nor does a claim of corporate aiding and abetting appear to meet
the requirement in Sosa that norms, to be actionable under the ATS,
must have at least the same “definite content and acceptance among
civilized nations [as] . . . the historical paradigms familiar when [the
ATS] was enacted.”304 There is little evidence that civil liability for
corporate aiding and abetting is widely accepted around the world.
While the concept of aiding and abetting liability is recognized as a
general matter in international criminal tribunals, it is applied in those
tribunals only to natural persons, not to corporations. Moreover, even
with respect to individuals in these cases, the standards for aiding and
abetting liability vary. For example, while the International Criminal
Tribunal for the Former Yugoslavia requires an aider or abettor to
have mere knowledge that his acts assist in a crime, the International
Criminal Court Statute is more demanding, requiring that the aider or
abettor act with the purpose of facilitating the commission of the
crime.305
A comparison between the claim rejected in Sosa and the argument
for imposing aiding and abetting liability on corporations is revealing.
The Court in Sosa rejected an arbitrary detention claim under the
ATS even though a norm prohibiting nations from arbitrarily detaining individuals was expressly included in the ICCPR, numerous other
treaties, the Restatement of Foreign Relations Law, and 119 national
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
302
303
See id. at 180–85, 190–91.
See id. at 181 (noting that “[t]he doctrine has been at best uncertain in application”); id. at
188 (noting that “the rules for determining aiding and abetting liability are unclear”).
304 Sosa, 124 S. Ct. at 2765.
305 Compare Statute of the International Criminal Tribunal for the Former Yugoslavia, supra
note 291, art. 7(1), with Rome Statute of the International Criminal Court, supra note 291, art.
25(3)(c). See also 1 THE ROME STATUTE OF THE INTERNATIONAL CRIMINAL COURT, supra
note 292, at 801. Some courts have expressed in dicta the view that a 1795 Attorney General
opinion, which the Court referred to in Sosa, provides support for aiding and abetting liability
under the ATS. See, e.g., Sarei v. Rio Tinto, PLC, 456 F.3d 1069, 1078 n.5 (9th Cir. 2006); Mujica
v. Occidental Petroleum Corp., 381 F. Supp. 2d 1164, 1173 n.6 (C.D. Cal. 2005). The 1795 opinion
observes that President Washington declared in his 1793 neutrality proclamation that individuals
“committing, aiding, or abetting hostilities” would “render themselves liable to punishment under
the laws of nations.” 1 Op. Att’y Gen. 57, 59 (1795) (opinion of William Bradford). In Sosa, the
Court cited to different language in this opinion that specifically referred to jurisdiction under the
ATS as support for the proposition that some common law causes of action could historically be
brought under the ATS. See Sosa, 124 S. Ct. at 2759 (quoting 1 Op. Att’y Gen. at 59). The aiding
and abetting language in the 1795 opinion, however, was not referring to the ATS or even to civil
liability; rather, it was referring to potential criminal liability. Moreover, the opinion obviously
provides no evidence that aiding and abetting liability is currently an accepted international law
norm in the civil context.
928
HARVARD LAW REVIEW
[Vol. 120:869
constitutions.306 The gap between international aspiration and enforceable ATS claims that was too large in Sosa is significantly larger
with respect to corporate aiding and abetting liability for human rights
abuses. There is no relevant treaty that embraces aiding and abetting
liability for corporations, the Restatement says nothing about such liability, and there is no widespread state practice of imposing liability
on corporations for violations of international human rights law. To
paraphrase Sosa, that a rule of corporate liability is so far from full realization is evidence against its status as binding law and even stronger
evidence against the creation by judges of a private cause of action to
enforce the aspiration behind the rule.307
The “practical considerations” adverted to by the Court in Sosa
also weigh against judicial recognition of corporate aiding and abetting
liability. These suits entail assessments of foreign government conduct
that is otherwise immune from U.S. jurisdiction under the Foreign
Sovereign Immunities Act. They also pose a risk of interfering with
political branch management of U.S. relations with the relevant countries — for example, in choosing whether to promote or restrict investment in these countries.308 And this litigation is likely to be perceived as improperly extraterritorial, especially when directed at
foreign companies.309 Invoking these policy concerns, the executive
branch has expressly opposed corporate aiding and abetting liability
under the ATS.310 Consistent with Sosa (and Erie), assessment of such
policy issues is best left to the political branches.
Finally, the Court in Sosa made two specific references to corporate
ATS litigation, and neither reference was supportive of aiding and
abetting liability. The Court stated in a footnote that, in considering
whether a norm is sufficiently definite to support a cause of action under the ATS, “[a] related consideration is whether international law extends the scope of liability for a violation of a given norm to the perpetrator being sued, if the defendant is a private actor such as a
corporation or individual.”311 Although cryptic, this statement suggests that courts should not broaden the “scope of liability” under international law through concepts such as aiding and abetting. The
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
306
307
308
See supra pp. 897–900.
See Sosa, 124 S. Ct. at 2769 n.29.
Cf. Crosby v. Nat’l Foreign Trade Council, 530 U.S. 363, 374 (2000) (noting that Congress
thought the Executive needed flexibility in managing sanctions and incentives aimed at improving human rights in Burma).
309 See, e.g., Brief for the Gov’ts of the Commonwealth of Austrl., the Swiss Confederation, &
the U.K. of Gr. Brit. & N. Ir. as Amici Curiae Supporting Petitioner, Sosa, 124 S. Ct. 2739 (No.
03-339).
310 See, e.g., Supplemental Brief for the U.S. of Am. as Amicus Curiae Supporting Appellants,
Doe I v. Unocal Corp., 395 F.3d 932 (9th Cir. 2002) (Nos. 00-56603, 00-56628).
311 Sosa, 124 S. Ct. at 2765 n.20 (emphasis added).
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
929
Court also referred at length in another footnote to a pending ATS
case brought against corporations that had done business with South
Africa during the apartheid regime and wrote that there was a “strong
argument” that courts should defer to the executive branch’s view that
this litigation would interfere with U.S. foreign relations.312 These
statements, along with the more general points discussed earlier, suggest that corporate aiding and abetting liability is improper under the
ATS after Sosa. Whether corporations should be liable for aiding and
abetting violations of customary international law is an issue that will
need to be addressed in the first instance by the political branches.
2. The War on Terrorism. — In the wake of the September 11 attacks and the ensuing “war on terrorism,” many of the alleged enemy
combatants in U.S. custody have, in various ways, invoked CIL as
federal law that, in their view, limits the Executive’s discretion to conduct the war. Detainees at Guantanamo have argued, for example,
that even if they are not directly covered by the Geneva Conventions,
the standards reflected in Common Article 3 of the Conventions are
binding on the United States as a matter of CIL and that these standards preclude trial by military commission.313 They have also argued
that their ongoing detention violates CIL prohibitions on arbitrary and
prolonged detention that bind the President as part of U.S. federal
common law314 and have sought remedies for interrogation techniques
and conditions of confinement that allegedly violate CIL norms.315
Individuals allegedly subject to detention or rendition elsewhere have
likewise asserted violations of CIL norms against prolonged arbitrary
detention, as well as torture and other cruel, inhuman, and degrading
treatment.316
It is highly unlikely that such claims can be brought against the
government in an ATS suit after Sosa. As an initial matter, the U.S.
government is presumptively immune from suit in U.S. courts. The
Federal Tort Claims Act317 (FTCA) partially waives sovereign immu–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
312
313
Id. at 2766 n.21.
See, e.g., Brief for Petitioner Salim Ahmed Hamdan at 48–50, Hamdan v. Rumsfeld, 126 S.
Ct. 2749 (2006) (No. 05-184), 2006 WL 53988. Common Article 3 prohibits, among other things,
“the passing of sentences and the carrying out of executions without previous judgment pronounced by a regularly constituted court affording all the judicial guarantees which are recognized as indispensable by civilized peoples.” Geneva Convention Relative to the Treatment of
Prisoners of War art. 3, opened for signature Aug. 12, 1949, 6 U.S.T. 3316, 75 U.N.T.S. 135.
314 See, e.g., In re Guantanamo Detainee Cases, 355 F. Supp. 2d 443, 445, 453 (D.D.C. 2005);
Khalid v. Bush, 355 F. Supp. 2d 311, 316–17, 328–29 (D.D.C. 2005).
315 See, e.g., Rasul v. Rumsfeld, 414 F. Supp. 2d 26 (D.D.C. 2006).
316 See, e.g., Complaint at 20–25, El-Masri v. Tenet, 437 F. Supp. 2d 530 (E.D. Va. 2006) (No.
1:05-cv-01417); Petition for Writ of Habeas Corpus at 2, 16, Omar v. Harvey, 416 F. Supp. 2d 19
(D.D.C. 2006) (No. 1:05-cv-02374).
317 28 U.S.C. §§ 1346(b), 2401(b), 2671–2680 (2000).
930
HARVARD LAW REVIEW
[Vol. 120:869
nity, but it has an exception for claims “arising in a foreign country”318
— an exception that the Court in Sosa, in a part of the opinion not
discussed in detail earlier in this Article, construed in favor of the government.319 This and related immunity doctrines impose a significant
obstacle to ATS suits against the U.S. government and its officials.320
Even if the immunity obstacle could be overcome, any ATS claim
against the government would need to satisfy the requirements imposed by Sosa, including the requirement that the CIL norms in question be widely accepted and specifically defined. Post-Sosa, it is difficult to say in the abstract whether any given norm would satisfy these
requirements. The Court in Sosa also made clear that, in deciding
whether to allow a CIL claim, courts should be “particularly wary of
impinging on the discretion of the Legislative and Executive Branches
in managing foreign affairs.”321 This separation of powers consideration is especially strong with respect to claims directed at the executive
branch’s management of a war.
War-on-terror claims brought outside the ATS raise additional issues. One fundamental issue is whether courts can apply CIL to override presidential action in the absence of some affirmative authorization in a treaty or statute. Whether CIL binds the President as a
matter of domestic law has been the subject of significant academic
debate.322 In The Paquete Habana, the Supreme Court stated that it
was appropriate to apply CIL “where there is no treaty, and no controlling executive or legislative act or judicial decision.”323 In light of
this statement, most lower courts have held that CIL cannot be applied to override the “controlling executive acts” of the President and
other high-level executive officials.324 Although Sosa did not address
the precise issue, its implicit rejection of the modern position, de–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
318
319
320
28 U.S.C. § 2680(k) (2000).
See Sosa v. Alvarez-Machain, 124 S. Ct. 2739, 2749–54 (2004).
See CURTIS A. BRADLEY & JACK L. GOLDSMITH, FOREIGN RELATIONS LAW 534–35
(2d ed. 2006).
321 Sosa, 124 S. Ct. at 2762.
322 See Essays, Agora: May the President Violate Customary International Law? (pts. 1 & 2), 80
AM. J. INT’L L. 913 (1986), 81 AM. J. INT’L L. 371 (1987); Glennon, supra note 10; Panel Session,
The Authority of the United States Executive To Interpret, Articulate or Violate the Norms of International Law, 80 AM. SOC’Y INT’L L. PROC. 297 (1986); Weisburd, supra note 73.
323 The Paquete Habana, 175 U.S. 677, 700 (1900); see also William S. Dodge, The Story of The
Paquete Habana: Customary International Law as Part of Our Law, in INTERNATIONAL LAW
STORIES (Dickinson et al. eds., forthcoming 2007) (manuscript at 18–22, on file with the Harvard
Law School Library) (arguing that The Paquete Habana should not be read to suggest that the
President can unilaterally disregard CIL).
324 See, e.g., Barrera-Echavarria v. Rison, 44 F.3d 1441, 1451 (9th Cir. 1995); Gisbert v. U.S.
Attorney Gen., 988 F.2d 1437, 1448 (5th Cir. 1993); Garcia-Mir v. Meese, 788 F.2d 1446, 1454–55
(11th Cir. 1986). But see In re Agent Orange Prod. Liab. Litig., 373 F. Supp. 2d 7, 109–10
(E.D.N.Y. 2005) (reasoning that CIL binds the President at least absent an official presidential
proclamation to the contrary).
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
931
scribed earlier, would seem to preclude binding the President to CIL as
a matter of domestic law in the absence of an incorporating statute or
treaty. If CIL is not automatically domestic federal law, then it is hard
to see how it is binding on the President as part of the “Laws” that he
must faithfully execute under Article II.
Another likely obstacle for war-on-terror claims brought outside the
ATS is the lack of congressional authorization. Even in the context of
a claim against a private foreign citizen, the Court in Sosa searched for
congressional authorization for the application of CIL. It is difficult to
find any congressional authorization, however, for the judicial application of CIL to regulate the war on terrorism. For example, following
September 11, Congress passed the Authorization for Use of Military
Force325 (AUMF), which broadly authorized the President to use “all
necessary and appropriate force” against al Qaeda and related entities,326 but did not refer to CIL, let alone domestic court application of
CIL, in its authorization. While the customary laws of war may inform the powers that Congress has implicitly conferred on the President in the AUMF, there is no suggestion that Congress intended to
impose affirmative CIL constraints on the President, much less judicially enforceable CIL constraints.327
The need for courts to find congressional authorization to apply international law to the war on terrorism is illustrated by the Supreme
Court’s decision in Hamdan v. Rumsfeld.328 In Hamdan, the Court
held that the military commissions that President Bush had established
after the September 11 attacks were not properly constituted because,
among other things, they failed to comply with requirements in Common Article 3 of the Geneva Conventions.329 Importantly, however,
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
325
326
327
Pub. L. No. 107-40, 115 Stat. 224 (2001) (to be codified at 50 U.S.C. § 1541 note).
Id. § 2(a), 115 Stat. at 224.
See Curtis A. Bradley & Jack L. Goldsmith, Congressional Authorization and the War on
Terrorism, 118 HARV. L. REV. 2047, 2091–2100 (2005). More recently, in the Detainee Treatment
Act of 2005, Pub. L. No. 109-148, div. A, tit. X, 119 Stat. 2739 (to be codified in scattered sections
of 10, 28, and 42 U.S.C.), Congress prohibited “cruel, inhuman, or degrading treatment or punishment” of anyone “in the custody or under the physical control of the United States Government.” Id. § 1003(a), 119 Stat. at 2739. This statute purported to incorporate a treaty obligation,
not a CIL obligation. In addition, Congress did not provide an enforcement mechanism for the
prohibition and in the same statute appeared to preclude or at least limit the Guantanamo detainees’ ability to raise this and other treatment-related claims in U.S. courts. See id. § 1005(e), 119
Stat. at 2742 (to be codified at 28 U.S.C. § 2241) (authorizing the D.C. Circuit to evaluate
“whether the status determination of the Combatant Status Review Tribunal with regard to [a
current detainee] . . . was consistent with the standards and procedures specified by the Secretary
of Defense” and whether those procedures and standards are consistent with any applicable provisions of the U.S. Constitution and laws, but eliminating both habeas corpus review for detainees
and jurisdiction over “any other action against the United States or its agents” by a current detainee or a former detainee who was “properly detained as an enemy combatant”).
328 126 S. Ct. 2749 (2006).
329 See id. at 2793–97.
932
HARVARD LAW REVIEW
[Vol. 120:869
the Court repeatedly emphasized that it was applying these requirements because they had been incorporated into U.S. domestic law by
Congress. The Court assumed for the sake of argument that Common
Article 3 could not be invoked “as an independent source of law”330
but reasoned that it was nevertheless part of the international “laws of
war” and that Congress in § 821 of the Uniform Code of Military Justice had required the President to comply with the laws of war in establishing military commissions.331 Justice Kennedy’s concurrence further emphasized this congressional incorporation of Common Article
3.332 This insistence on congressional authorization for domestic court
application of a treaty provision that had already been expressly ratified by the political branches suggests, a fortiori, that there is such a
requirement for domestic court application of the unwritten norms of
CIL, which may arise internationally without the express endorsement
of the political branches.
None of the points made thus far implies that the United States
lacks an international obligation to comply with norms of CIL relevant to the war on terrorism or that the political branches should not
take account of those obligations in conducting the war. Even when
CIL is not enforceable by U.S. courts, it still binds the United States
on the international plane. This point was obscured in an early draft
of an Office of Legal Counsel memorandum concerning the applicability of the Geneva Conventions to the war on terrorism, in which the
authors asserted that “any customary international law of armed conflict in no way binds, as a legal matter, the President or the U.S.
Armed Forces concerning the detention or trial of members of al
Qaeda and the Taliban.”333 This assertion is true, at most, only with
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
330
331
Id. at 2794.
See id. at 2774, 2786, 2794. At the time of this decision, § 821 provided that “the provisions
of this chapter conferring jurisdiction upon courts-martial do not deprive military commissions . . . of concurrent jurisdiction with respect to offenders or offenses that by statute or by the
law of war may be tried by military commissions.” 10 U.S.C.A. § 821 (West 1998 & Supp. 2006)
(emphasis added).
332 See Hamdan, 126 S. Ct. at 2799 (Kennedy, J., concurring) (“[This] is a case where Congress,
in the proper exercise of its powers as an independent branch of government, and as part of a
long tradition of legislative involvement in matters of military justice, has considered the subject
of military tribunals and set limits on the President’s authority.”); id. (“[T]he requirement of the
Geneva Conventions of 1949 that military tribunals be ‘regularly constituted’ . . . controls here, if
for no other reason, because Congress requires that military commissions like the ones at issue
conform to the ‘law of war.’” (citation omitted)); id. at 2802 (“Common Article 3 is part of the law
of war that Congress has directed the President to follow in establishing military commissions.”).
333 Draft Memorandum from John Yoo, Deputy Assistant Attorney Gen., and Robert J. Delahunty, Special Counsel, U.S. Dep’t of Justice, to William J. Haynes II, Gen. Counsel, U.S. Dep’t
of Def. 34 (Jan. 9, 2002), available at http://www.gwu.edu/~nsarchiv/NSAEBB/NSAEBB127/
02.01.09.pdf.
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
933
respect to domestic law, not international law. The final version of the
memorandum properly refined this assertion.334
3. International and Foreign Sources in Constitutional Interpretation. — In recent years, the Supreme Court has cited and relied on international and foreign materials in the course of interpreting proviThis practice has generated
sions of the U.S. Constitution.335
significant controversy in the academy, among policymakers, and
among members of the Court.336 There has been little discussion,
however, of the relationship between this practice and the practice of
applying CIL as domestic law.337
We begin with the relationship between an internationalized constitutionalism and the modern position. The two practices bear certain
similarities. Both modern position advocates and those advocating an
internationalized constitutionalism invoke the same basic sources —
treaties (sometimes non-self-executing or unratified ones), foreign laws
and decisions, U.N. resolutions, the writings of jurists, and the like —
in an effort to persuade domestic courts to grant relief not otherwise
available under U.S. law. Moreover, the modern position and internationalized U.S. constitutionalism are complementary strategies for
achieving domestic legal change. A good example of this is the juvenile death penalty. For years, litigants argued, largely unsuccessfully,
that an alleged CIL prohibition on the execution of juvenile offenders
was binding domestic law that preempted state juvenile death penalty
laws.338 These litigants were eventually more successful, however, in
using nearly identical sources to convince the Supreme Court that the
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
334 See Memorandum from Jay S. Bybee, Assistant Attorney Gen., U.S. Dep’t of Justice, to Alberto R. Gonzales, Counsel to the President, and William J. Haynes II, Gen. Counsel, U.S. Dep’t
of Def. 32 (Jan. 22, 2002), available at http://www.gwu.edu/~nsarchiv/NSAEBB/NSAEBB127/
02.01.22.pdf (“Customary international law . . . cannot bind the executive branch under the Constitution because it is not federal law.” (emphasis added)).
335 See, e.g., Roper v. Simmons, 125 S. Ct. 1183, 1198–1200 (2005); Lawrence v. Texas, 539 U.S.
558, 572–73 (2003); Atkins v. Virginia, 536 U.S. 304, 316 n.21 (2002).
336 For articles supporting this practice, see, for example, Sarah H. Cleveland, Our International Constitution, 31 YALE J. INT’L L. 1 (2006); Vicki C. Jackson, Constitutional Comparisons:
Convergence, Resistance, Engagement, 119 HARV. L. REV. 109 (2005); Harold Hongju Koh, International Law as Part of Our Law, 98 AM. J. INT’L L. 43 (2004); Gerald L. Neuman, The Uses of
International Law in Constitutional Interpretation, 98 AM. J. INT’L L. 82 (2004); and Jeremy
Waldron, Foreign Law and the Modern Ius Gentium, 119 HARV. L. REV. 129 (2005). For articles
critical of this practice, see, for example, Roger P. Alford, Misusing International Sources To Interpret the Constitution, 98 AM. J. INT’L L. 57 (2004); Robert J. Delahunty & John Yoo, Against
Foreign Law, 29 HARV. J.L. & PUB. POL’Y 291 (2005); John O. McGinnis, Foreign to Our Constitution, 100 NW. U. L. REV. 303 (2006); and Ernest A. Young, Foreign Law and the Denominator
Problem, 119 HARV. L. REV. 148 (2005). For examples of controversy among policymakers, see
Cleveland, supra, at 4 & nn.14–19.
337 A partial exception is Professor Waldron, who discusses Sosa and Roper together but does
not analyze their relationship. See Waldron, supra note 336.
338 For examples, see Curtis A. Bradley, The Juvenile Death Penalty and International Law, 52
DUKE L.J. 485, 490 n.14 (2002).
934
HARVARD LAW REVIEW
[Vol. 120:869
Eighth Amendment, interpreted in light of these sources, prohibited
the execution of juvenile offenders.339
Despite these similarities, there are significant differences between
the modern position and the use of international and foreign materials
in constitutional interpretation. From one perspective, the use of international and foreign materials in constitutional interpretation raises
more significant normative concerns than the modern position. While
Congress can overrule any judicial domestication of CIL, a point emphasized in Sosa,340 constitutional interpretations bind Congress and
can be overturned only through a constitutional amendment. Thus,
the use of international and foreign materials in constitutional interpretation raises two levels of potential antimajoritarian concern: unelected federal judges incorporate foreign materials into U.S. law, and
they do so in a way that permanently displaces the political branches
from their usual role in this regard.
Whatever their similarities, an internationalized constitutionalism
does not entail or even support the modern position that all of CIL is
domestic federal law. Courts have drawn on foreign and international
sources in interpreting the Constitution throughout U.S. history, including during the first 150 years of the nation when CIL clearly did
not have the status of domestic federal common law.341 Moreover, the
Supreme Court’s constitutional decisions drawing on foreign and international sources have treated these sources, at most, as potentially
relevant to the interpretation of vague or uncertain constitutional provisions, not as sources of law that have direct and binding application
in the U.S. legal system. The Court has emphasized, for example, that
“[t]he opinion of the world community, while not controlling our outcome, does provide respected and significant confirmation for our own
conclusions.”342 By contrast, under the modern position, CIL is not
merely an interpretive tool but is binding of its own force in U.S.
courts in a way that is not tethered to any extant federal law.
When we compare the trend towards internationalized constitutionalism with the Supreme Court’s analysis in Sosa, further contrasts
appear. The Court has been much less rigorous with respect to foreign
and international materials in its constitutional interpretation cases
than it was with respect to these sources in the context of the ATS in
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
339
340
341
342
See Roper, 125 U.S. at 1198–1200.
See Sosa v. Alvarez-Machain, 124 S. Ct. 2739, 2765 (2004).
For examples, see Cleveland, supra note 336; and Jackson, supra note 336, at 109–11.
Roper, 125 S. Ct. at 1200 (emphasis added). Some might think that internationalized constitutionalism is akin to the interpretive use of CIL under the Charming Betsy canon of construction. Cf. Daniel Bodansky, The Use of International Sources in Constitutional Opinion, 32 GA. J.
INT’L & COMP. L. 421 (2004) (arguing that the Charming Betsy canon should be applied to constitutional interpretation). For arguments to the contrary, see Bradley, supra note 338, at 555–56;
McGinnis, supra note 336, at 307 n.23.
2007]
SOSA, CUSTOMARY INTERNATIONAL LAW, AND ERIE
935
Sosa. In Roper v. Simmons,343 for example, in which the Court held
that the execution of juvenile offenders violates the Eighth Amendment, the Court cited, among other things, the Convention on the
Rights of the Child,344 a treaty that had not been ratified by the
United States, and the ICCPR, which the U.S. had ratified with a reservation declining to agree to the ban in that treaty on the juvenile
death penalty.345 By contrast, in Sosa, as we discussed earlier, the
Court described the ICCPR as having “little utility” in its analysis,
even though, unlike in Roper, there was no relevant reservation with
respect to the issue before the Court.346
It is difficult to know what to make of the Supreme Court’s differing treatment of foreign and international sources in the constitutional
and ATS contexts. The application of foreign law in both contexts
might be viewed as consistent with Erie’s positivism because in both
contexts the Court relies on a domestic sovereign source that purportedly makes relevant the foreign and international materials, and because the resulting legal conclusions reflect domestic U.S. law.347 Nevertheless, the Court has a more developed theory of the relevance of
foreign and international law sources in the ATS context than in the
constitutional context — a theory that, consistent with Erie, severely
limits judicial discretion in relying on foreign and international
sources. If the Court begins to place more significant weight on these
materials in its constitutional decisions, it will need to pay greater attention to the limitations of these materials, just as it did in the ATS
context in Sosa.
VI. CONCLUSION
The Supreme Court’s decision in Sosa resolved a number of the
debates concerning the domestic status of CIL. The Court confirmed
that CIL historically had the status of nonfederal general law. The
Court also made clear that any evaluation of CIL’s modern status must
operate against the background of Erie and the limitations of the postErie federal common law. Most importantly, the Court’s reasoning
and conclusions are incompatible with the modern position claim that
CIL is automatically part of U.S. federal law. CIL is incorporated into
federal law, under the analysis in Sosa, only when its incorporation has
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
343
344
345
346
347
125 S. Ct. 1183.
Opened for signature Nov. 20, 1989, 1577 U.N.T.S. 3.
See Roper, 125 S. Ct. at 1199.
See supra pp. 899.
Cf. Waldron, supra note 336, at 143 (noting that in its role of informing the development of
domestic law, “it is not necessary that ius gentium be understood positivistically; it need only be
seen as a source of normative insight grounded in the positive law of various countries and relevant to the solution of legal problems in this country”).
936
HARVARD LAW REVIEW
[Vol. 120:869
been authorized either by the structure of the Constitution or by the
political branches, and it is to be applied interstitially in a manner consistent with the relevant policies of the political branches. Nevertheless, because there are a number of plausible structural and statutory
authorizations for the domestication of CIL in select areas, this body of
international law will continue to play an important role in U.S. judicial decisionmaking and therefore will continue to be, in the words of
The Paquete Habana, “part of our law.”